Making friends with benefits: an investigation into the use of

Making friends with benefits: an investigation into the use of
incentives for the conservation of forest commons
Henry Travers
Department of Life Sciences, Imperial College London
A dissertation submitted for the degree of Doctor of Philosophy
Imperial College London
May 2014
1
Declaration
This dissertation is the result of my own work and includes nothing which is the
outcome of work done by or in collaboration with others, except where specifically
indicated in the text.
Henry Travers, May 2014
The copyright of this thesis rests with the author and is made available under a
Creative Commons Attribution Non-Commercial No Derivatives licence. Researchers
are free to copy, distribute or transmit the thesis on the condition that they attribute it,
that they do not use it for commercial purposes and that they do not alter, transform or
build upon it. For any reuse or redistribution, researchers must make clear to others
the licence terms of this work
2
Abstract
The use of incentives to encourage sustainable natural resource use is widespread in
conservation. Yet, the relative or combined effects of conservation interventions on
resource use can be difficult to predict. This is particularly so in contexts with weak
institutions or poorly defined property rights, conditions that typify forest commons.
In this study, I investigate the effect of different policy options, currently being
implemented or under consideration in a protected area in Cambodia, on the resource
appropriation of indigenous smallholder farmers in order to develop a framework for
predicting the effectiveness of conservation interventions.
I begin by evaluating the progress made towards two project targets relating to the
security of natural resources important for local livelihoods, showing that, whilst
security issues exist, the project has so far been successful in protecting key resources.
I also evaluate an existing programme of indigenous land titling, which is shown to be
consistent with conservation goals but vulnerable to interference from land grabs.
Using two approaches for predicting behavioural response to planned interventions,
experimental games and scenario analysis, I demonstrate that collective performance
payments are more effective than individually contracted payments or increased law
enforcement effort. Previous research has shown that externally imposed rule
enforcement can undermine existing norms for cooperation, particularly in contexts
where social cohesion is high. I build upon this research by demonstrating that
exogenous rule enforcement, when coupled with reward payments that have been
found to encourage high levels of cooperation, can increase resource extraction
compared to when payments are offered in isolation.
The research presented highlights the difficulty of predicting the effect of policies
designed to reduce resource use, particularly when interventions interact to increase
the institutional complexity in which resource use decision-making takes place, but
offers a novel framework for the investigation of intervention effectiveness.
3
Acknowledgements
I am hugely grateful to my two supervisors, E.J. Milner-Gulland and Tom Clements.
They have both advised, encouraged and inspired me in equal measure. Put simply, I
would not have made it this far without them. Thank you, both of you!
I will also always be indebted to Gareth Edwards-Jones.
So many people at the WCS Cambodia Program have helped me along the way. I
would like to thank them all, but particularly Ashish John for challenging me and
making me think, Pet Phaktra for allowing me to benefit from his wealth of
experience, Phien Sayon for all his help negotiating labyrinthine GIS archives, Tim
Solita for her amazing patience and, not least, Tao Sarath and Khuth Channy for
putting up with my unconventional accounting. Special thanks go to Tom Evans. His
guidance, advice and patience under a barrage of questions have been invaluable.
I could not have completed the fieldwork aspects of this study without the hard work
and commitment of Kyle Winney, Yu Kimleng, Orn Samart, Heng Hong, Chum Tevi,
Mao Tineul, and Lou Vanny.
The ICCS group at Silwood have been fantastic sounding boards, rivals and
supporters. Tim, Joe and, lately, Ben, you have made excellent cohort mates and I feel
no shame in ending up with the wooden spoon.
To all my family and friends, I am sorry to have inflicted the last three and a half
years on you. Thank you all for your unending patience and support. Mum – I hope
that one day you can accept that science does not always have to wear a white coat!
Un grande grazie Tiziana, per il tuo amore e ottimismo infinito.
This work was supported by the Economic and Social Research Council and the
Natural Environment Research Council.
4
Abbreviations and Acronyms
AIC
Akike Information Criterion
AICc
corrected Akike Information Criterion
BLUPs
Best Linear Unbiased Predictors
CBD
Convention of Biological Diversity
CBNRM
Community Based Natural Resource Management
CCB
Carbon, Community and Biodiversity
CCBA
Carbon, Community and Biodiversity Association
CEDAC
Cambodian Centre for Development and Study in Agriculture
CENTDOR
Centre for Development Orientated Research in Agriculture and
Livelihood Systems
CPR
Common Pool Resource
ELC
Economic Land Concession
FA
Forestry Administration
GPS
Global Positioning System
GLM
Generalised Linear Model
GLMM
Generalised Linear Mixed Model
IC
Indigenous Community
ICCA
Indigenous or Community Controlled Area
ICDP
Integrated Conservation and Development Project
ICT
Indigenous Communal Title
IP
Indigenous People
LMM
Linear Mixed Model
MoE
Ministry of Environment
NGO
Non-Governmental Organisation
NTFP
Non-Timber Forest Product
PES
Payments for Environmental Services
REDD+
Reduced Emissions from Deforestation and forest Degradation
SD
Standard Deviation
SE
Standard Error
SPF
Seima Protection Forest
ToC
Theory of Change
UNFCCC
United Nations Framework Convention on Climate Change
VCS
Verified Carbon Standard
WCS
Wildlife Conservation Society
5
Table of Contents
Introduction ...................................................................................................................... 11
Problem statement ........................................................................................................11
Aims and objectives .......................................................................................................14
Thesis outline ..................................................................................................................14
Seima Protection Forest ................................................................................................ 17
Introduction .....................................................................................................................17
Biophysical characteristics .........................................................................................18
Human geography ..........................................................................................................18
Site management ............................................................................................................21
Assessment of progress towards social targets of the Seima Protection
Forest conservation programme ............................................................................... 25
Introduction .....................................................................................................................25
Methods .............................................................................................................................29
Results ................................................................................................................................40
Discussion .........................................................................................................................54
A tale of two villages: an investigation of conservation-driven land tenure
reform in a Cambodian Protection Forest .............................................................. 59
Introduction .....................................................................................................................59
Indigenous land tenure reform in Cambodia .......................................................61
Land tenure reform in the two villages ..................................................................63
Methods .............................................................................................................................66
Results ................................................................................................................................70
Discussion .........................................................................................................................79
Applying the carrot and stick in a Cambodian commons: an experimental
games approach to the investigation of conservation incentives .................. 84
Introduction .....................................................................................................................84
A common pool resource game .................................................................................87
Theoretical prediction and hypotheses .................................................................91
Methodology.....................................................................................................................93
Results ................................................................................................................................95
Discussion ...................................................................................................................... 100
6
Investigating incentive based approaches to biodiversity conservation
through scenarios ..........................................................................................................105
Introduction .................................................................................................................. 105
The Seima Protection Forest REDD+ Demonstration Project ..................... 108
Methods .......................................................................................................................... 110
Results ............................................................................................................................. 114
Discussion ...................................................................................................................... 123
Discussion ........................................................................................................................127
Benefit sharing for REDD+ ....................................................................................... 127
Experimental games vs. scenario interviews .................................................... 130
Complexity in conservation policy research ..................................................... 133
Opportunities for future research ........................................................................ 135
Appendix A: Supplementary material for Chapter 3 .........................................153
Appendix B: Supplementary material for Chapter 4 .........................................155
Appendix C: Supplementary material for Chapter 5 .........................................162
Appendix D: Supplementary material for Chapter 6 ........................................187
7
List of Tables
Table 3.1: Selected indicators for the two social management targets identified for Seima
Protection Forest. The expected trend that would indicate progress towards project targets is
given for each indicator.
30
Table 3.2: Hypothesised vulnerable and special interest groups identified for the impact
assessment. These groups are not exclusive and so it was possible for households to be classified
as belonging to more than one group.
34
Table 3.3: Coefficient estimates for the LMM with the full household poverty score for 2012 as
the response variable. Standard errors are given in parentheses. R2m = 0.363, R2c = 0.416. Resin
tree ownership at the intercept was taken to be 0-50 trees.
42
Table 3.4: Coefficient estimates for the LMMs with the reduced household poverty score as the
response variables. Standard errors are given in parentheses.R2m = 0.330, R2c = 0.335.
45
Table 3.5: Coefficient estimates for the GLMM with the probability of owning resin trees. The
difference in probability was found by a reverse transformation of the logit link function used in
the GLMM for households of sample average size. Standard errors are given in parentheses.
Significance values: ** = P < 0.01; *** = P < 0.001.
47
Table 3.6: Coefficient estimates for LMMs, with the natural logarithm of trees owned by resin
tree owning households as the response variable. Standard errors are given in parentheses. R2m =
0.034, R2c = 0.149.
48
Table 3.7: Coefficient estimates for the GLMM with the number of wild meat meals consumed
per household in one week in 2012 as the response variable. The difference in the number of wild
meals eaten was found by a reverse transformation of the logit link function used in the GLMM
for an average household (i.e. assuming sample average reduced poverty score, household size,
education level and distance to nearest market). The difference in number of wild meat meals
consumed provides an estimate of the effect of belonging to each of the different socio-economic
groups included in the model. Standard errors are given in parentheses. Significance values: ** =
P < 0.01; *** = P < 0.001.
49
Table 3.8: Coefficient estimates for the GLMM with the probability of feeling secure about
natural resources important for livelihoods. The difference in probability was found by a reverse
transformation of the logit link function used in the GLMM for an average household (i.e.
assuming sample average number of resin trees per household). Standard errors are given in
parentheses. Significance values: * = P < 0.05; ** = P < 0.01; *** = P < 0.001.
51
Table 3.9: Coefficient estimates for LMM with the log transformed reported area of land claimed
per household as the response variable. Standard errors are given in parentheses. R2m = 0.175,
R2c = 0.216.
52
Table 3.10: Coefficient estimates for the GLMM with the probability of a household being
landless as the response variable. The difference in probability was found by a reverse
transformation of the logit link function used in the GLMM for an average household (i.e.
assuming sample average household poverty score). Standard errors are given in parentheses.
Significance values: ns = P > 0.05; *** = P < 0.001.
53
Table 4.1: Details of the participatory land use planning process in the two study villages. Land
inside areas designated for agriculture is a mix of cultivated land and forest that, as yet, has not
been cleared.
65
Table 4.2: Mean area [ha] given to different land uses per household for the two villages at the
time fields were mapped by PA authorities as part of the ICT process and in 2012. The
8
percentage of average household area given to each land use is shown in parentheses. As
intercropping is common practice, percentages do not sum to 100.
71
Table 4.3: Parameter estimates for the household land holdings LMM, with ln(area) as the
response variable. One random effect was included in the model: settlement. The full list of
variables modelled, plus their explanations, are given in Appendix B; Table B.2. Standard errors
are given in parentheses. R2m = 0.382, R2c = 0.410.
73
Table 4.4: Parameter estimates for the illegal household land holdings, with ln(illegal area) as the
response variable. P values significant at the 95% confidence level are shown in bold. The full list
of variables modelled, plus their explanations, are given in Appendix B; Table B.2. Standard
errors are given in parentheses. R2 = 0.613. Significance values: * = P < 0.05; *** = P < 0.001. 75
Table 4.5: Percentage of respondent households that were found to have complied with the five
community land use regulations. Only land inside the community agricultural areas was
considered.
77
Table 5.1: Treatment options for the CPR game. All treatments were played by 80 participants
with the exception of the private treatment, which was played by 360 participants. Treatment
parameters are given for the payoff equation (Equation 5.1).
88
Table 5.2: Descriptive statistics for key explanatory variables. Sample standard errors are shown
in parentheses.
96
Table 5.3: Parameter estimates for the selected CPR game generalised linear mixed models, with
the number of fish harvested by an individual in a single round as the response variable. The
difference in the number of fish taken was found by calculating the expected number of fish
taken for each variable and comparing this with the number expected for the intercept. Standard
errors in terms of number of fish are shown in parentheses. A comparison of the effect of each
treatment for round 1 and round 4 is given in Supplementary Materials Table C.4. The full list of
variables modelled, plus their explanations, are given in Supplementary Materials Table C.2.
Significance values: ns = not significant; (*) = P < 0.1; * = P < 0.05; ** = P < 0.01; *** = P < 0.001.
Reference level treatment indicated by ref.
97
Table 6.1: Parameter estimates for the pro-conservation and clearance reduction models
considering conditional and unconditional responses. The difference in the probability of acting
in a pro-conservation manner or reducing clearance relative to the baseline scenario was found
by calculating the expected probability each variable and comparing this with the probability
expected for the intercept. The full list of variables modelled, plus their explanations, are given in
Appendix D; Table D.1. Significance values: ns = not significant; (*) = P < 0.1; * = P < 0.05; ** =
P < 0.01; *** = P < 0.001. Reference level treatment indicated by ref.
116
9
List of Figures
Figure 2.1: Location of Seima Protection Forest in Cambodia (from Evans et al., 2012a).
17
Figure 2.2: Landscape showing the location of the 17 key villages and 3 other user villages
participating with the project and identified for inclusion in the social monitoring framework
(from WCS, 2013).
23
Figure 3.1: Conceptual model developed by PA authorities for Seima Protection Forest REDD+
Demonstration Project (from WCS, 2013)
28
Figure 3.2: Distribution of full household poverty scores from the 2012 quantitative household
livelihood survey.
41
Figure 3.3: Estimated relationship between the full household poverty score and the total
reported area claimed per household. The vertical dashed line is drawn at the mean area of
productive land claimed per household.
43
Figure 3.4: BLUPs for the village random effect. The x-axis shows the effect of living in a
particular village in terms of the difference to the full household poverty score from the intercept.
Error bars show the 95% confidence interval for each village.
44
Figure 3.5: Comparison of the distribution of reduced household poverty scores between the 2007
and 2012 quantitative household livelihood surveys.
45
Figure 3.6: Reasons given for why respondent households felt secure about their resources
51
Figure 4.1: Map of Seima Protection Forest (2,926 km2), showing the location of the two study
villages.
64
Figure 4.2: Predicted probabilities of compliance with the 2001 Land Law for non-IC and IC
households. The vertical dashed line shows the average area claimed per household for the whole
sample.
75
Figure 4.3 Predicted probabilities of compliance with IC regulations. The vertical dashed line
shows the average area claimed per household for the whole sample.
77
Figure 6.1 Proportion of coded responses for the business as usual baseline scenario and each of
the six future scenarios. Conditional behaviour refers to behaviour that is contingent on
reciprocation.
115
Figure 6.2: Proportion of coded responses for the business as usual baseline scenario and each of
the six future scenarios stratified by livelihood zone (row 1) and household poverty score terciles
(row 2). Conditional behaviour refers to responses that are contingent on reciprocation.
122
10
Chapter 1
1. Introduction
Problem statement
Following growing recognition, from the 1980s onwards, that the conservation of
biodiversity is intrinsically linked to the well-being of people living in areas of
biological wealth (Milner-Gulland et al., 2014), a number of policy paradigms have
emerged in which the dominant logic rests on resolving trade-offs between the
interests of local people and those pursuing an agenda of conservation (Adams &
Hulme, 2001; Brown, 2002; Adams et al., 2004; Wells & McShane, 2004; Sunderland
et al., 2007; Chhatre & Agrawal, 2009; McShane et al., 2011; Persha et al., 2011).
Central to all of these approaches has been the desire to create or alter the level of
incentives to reduce the impact of those people whose actions threaten biodiversity,
whether through the enforcement of rules restricting resource use (Gibson et al., 2005;
Keane et al., 2008), facilitation of alternative, less-damaging sources of income
(Brandon, 1996; Salafsky & Wollenberg, 2000), strengthening of property, access or
management rights (Singleton, 2000; Carlsson & Berkes, 2005) or, more recently,
direct payments or payments for the provision of environmental services (Ferraro &
Kiss, 2002; Pagiola et al., 2004; Wunder, 2007; Ferraro, 2011). Whilst all of these
approaches seek to change the incentives governing behaviour, their success is reliant
upon the institutional context in which they are applied (Barrett et al., 2001; Brown,
2003; Clements et al., 2010). There is, however, currently a paucity of research aimed
towards increasing understanding of how to facilitate the establishment of effective
conservation institutions, the “rules of the game” which govern human interactions
(North, 1990), particularly in the face of social, economic and environmental change,
or towards influencing behavioural change in order to achieve positive outcomes for
conservation (Cowling, 2014).
In recent years, increasing attention has been given to evidence-based approaches to
improving conservation decision-making and to evaluating whether or not
conservation policies have been successful in achieving their stated goals (Pullin &
Knight, 2003; Sutherland et al., 2004; Stem et al., 2005; Ferraro & Pattanayak, 2006;
11
Pullin & Knight, 2009). Yet the success of conservation policies is difficult to define
or measure and may be highly dependent on the local social, political, economic and
institutional context, which can lead to significant uncertainty (Adams et al., 2003;
McShane et al., 2010). Despite recent advances, such as the application of quasiexperimental matching approaches (e.g. Andam et al., 2008; Andam et al., 2010;
Arriagada et al., 2012; Ferraro & Hannauer, 2014; Clements et al., in press; Clements
& Milner-Gulland, in press), impact assessments are difficult to do well, are
expensive (Agrawal, 2014) and are conducted after policies have been implemented.
As such, even if approaches, such as adaptive management (Salafsky et al., 2002;
Stem et al., 2005), have been put in place, opportunities for improvements, time or
goodwill may be lost in the interim. Consequently, the ability to estimate the
effectiveness of policies prior to implementation (either in absolute terms or relative
to alternative policy options), through approaches such as management strategy
evaluation (Bunnefeld et al., 2011; Milner-Gulland, 2012; Nuno, 2013), offers the
potential to get things right first time and guide policy decision-making to minimise
the risk of unnecessary policy failures.
These concerns are particularly relevant for the conservation of tropical forests where
current international policy developments are poised to provide substantial additional
funding in the next few years through initiatives to reduce emissions from
deforestation and forest degradation (REDD+; Ebling & Yasué, 2008; Miles & Kapos,
2008). Hitherto, the clearance of tropical forests has proved challenging to address
(Geist & Lambin, 2002; Chomitz et al., 2007). Between 1990 and 2010 alone global
forest resources were lost at a net rate of approximately 6.8 million hectares per year
(FAO, 2010). With the advent of REDD+ initiatives, there is renewed optimism in
some circles that efforts to protect forests can be successfully supported. Yet in order
for this to happen, it will be necessary to channel potential carbon revenue effectively
to deliver actual behavioural change by those responsible for deforestation (Brown et
al., 2008). This is especially difficult in the context of weak institutions (Barrett et al.,
2001) and unclear property rights (Hardin, 1968), conditions that typify forest
commons and for which resource extraction is most likely to be unsustainable.
The proximate causes of deforestation are highly complex and governed by wide
ranging factors, such as rising human populations, immigration, property rights,
12
commodity prices, market access, as well as the development and integration of new
technologies (Laurance, 1999; Lambin et al., 2001; Geist & Lambin, 2002; Chomitz
et al., 2007; DeFries et al., 2010). Due to the complex nature of many of these drivers
it is consequently difficult to draw conclusions on the wider importance of different
determinants of forest depletion (Scrieciu, 2007), and hence set policy. The evidence
with regard to the association between small-scale, rural households, increased market
access and deforestation is particularly contradictory. As rural households often have
little access to capital or alternative livelihood activities, expansion into marginal
open-access forested land is an attractive option (Barbier, 1997). Increased access to
capital does not, however, necessarily reduce the rate of deforestation, as it frequently
leads to investment requiring or facilitating land clearance (Wunder, 2001; Pendleton
& Howe, 2002). In other cases, the creation of new market opportunities can lead to a
transition away from agriculture as a primary source of income and reduce pressure
on forest resources. As such, it is doubtful that a direct causal link between poverty
and deforestation exists in the general case but the fact remains that poor rural
households are frequently direct agents of deforestation (Sunderlin et al., 2005).
Given this, it is important for policy initiatives, such as REDD+, to consider not only
the role of small-scale resource users in forest clearance but also the impact of forest
protection measures on rural livelihoods and behaviour (Brown et al., 2008). In areas
principally under threat from small-scale deforestation, decision-makers must develop
appropriate policy interventions, which seek to maximise the incentives for local
people to reduce clearance whilst adhering to the principle of doing no harm.
This study seeks to contribute to our understanding of these issues by concentrating
on the complex inter-relationships between human livelihoods, land-use management,
and the protection of forests for the purposes of climate change mitigation and
biodiversity conservation. This research is focussed on a single protected area,
managed for the protection of biodiversity and the enhancement of local livelihoods,
in Cambodia, a country with one of the highest rates of percentage forest cover loss
globally (Hansen et al., 2013). The study site is currently in the final stages of
accreditation as a voluntary REDD+ project and has been identified as one of two
national REDD+ demonstration sites. Hence, the findings of this study will contribute
directly to management decisions at the study site. Revenue generated by the REDD+
project will be used to fund site management interventions, including benefit sharing
13
with 20 villages located within or adjacent to the site that have signed agreements to
participate in the project. As one of the principal causes of deforestation within site
boundaries is the expansion of agricultural land by smallholder farmers from
participating villages (WCS, 2013), the success of the project hinges on the ability of
site managers to incentivise these farmers to halt forest clearance. As such, the site
provides a significant opportunity to study the potential impact of alternative policy
options for the prevention of smallholder forest clearance.
Aims and objectives
The overall aim of this study was to investigate the effectiveness of different forest
conservation policy interventions currently implemented or planned for Seima
Protection Forest in terms of delivering behavioural change amongst small-scale
resource users.
The objectives of the study were as follows:
1. Characterise household livelihood strategies and economic well-being and
assess progress towards project livelihood targets for the period 2007 to 2012.
2. Investigate household land use and rule compliance under existing
conservation management interventions for the period 2004 to 2012.
3. Elucidate individual expectations of changes in land use strategies under a
range of future scenarios of conservation interventions and market conditions.
4. Identify implications for incentive-based conservation policy making, and
make recommendations for intervention design at the study site.
Thesis outline
In addition to the introduction, the thesis is split into six further chapters:
Chapter 2 provides an overview of the project study site, Seima Protection Forest,
and the livelihood strategies and practices of the people who live there, as well as
summarising the history and proposed future of conservation management at the site.
Chapter 3 focuses on assessing the livelihoods and economic well-being of
households living within the study site and evaluates the progress made towards two
site management goals (ensuring household resource security and access to sufficient
14
agricultural land) during the period 2007 to 2012. Household economic well-being is
shown to have improved over this period, even among vulnerable groups, except in
the most remote villages which still have limited access to emerging markets. Partial
progress has been made in relation to the two management goals: ensuring household
resource security and sufficient access to agricultural land, with household resource
security and access to agricultural land largely stable. The theory of change based
approach, as applied here, is found to be vulnerable to data and indicator selection
issues.
The results presented in this chapter appear in:
Travers, H. & Evans, T. (2013) Development of a social impacts monitoring system
for the Seima Core Protection Forest REDD+ Demonstration Site. Wildlife
Conservation Society, Phnom Penh.
Chapter 4 investigates household land use and rule compliance in two villages under
one of the principal interventions implemented at the study site, a programme of
participatory land use planning and tenure reform. I show that participatory tenure
reform can be consistent with positive outcomes for conservation, but that this process
is vulnerable to disruption from external interests in the form of immigrant
households and powerful land speculators.
This chapter is under revision at Land Use Policy as:
Travers, H., Winney, K., Clements, T., Evans, T.D., Milner-Gulland, E.J. A tale of
two villages: an investigation of conservation-driven land tenure reform in a
Cambodian Protection Forest.
Chapter 5 applies a behavioural games approach to the investigation of resource user
behaviour in a commons dilemma under a series of experimental treatments designed
to mimic policy options under consideration for the study site. The interaction
between rewards and penalties is examined, showing the potential for weakly
enforced penalties to “crowd out” reductions in resource use observed when rewards
are offered in isolation.
15
This chapter is under second review at Ecological Economics as:
Travers, H., Clements, T., Milner-Gulland, E.J. Applying the carrot and stick in a
Cambodian commons: an experimental games approach to the investigation of
conservation incentives.
Chapter 6 uses a scenario-based approach to investigate future smallholder land
clearance within the study site under a series of potential policy options and market
conditions. Collective payments or payments into a communal fund in return for
households choosing to halt clearance were found to be most effective at reducing
clearance relative to the business as usual baseline scenario, provided that other
members of the village reciprocated.
Chapter 7 provides a discussion of the main findings described in preceding chapters
and identifies opportunities for further study.
16
Chapter 2
2. Seima Protection Forest
Introduction
The study was conducted in Seima Protection Forest (SPF), a protected area located in
the Eastern Plains Landscape of Cambodia, which covers an area of 2,927 km2 and is
located in north-eastern Cambodia, at the southern end of the endemic-rich Annamite
Mountains (Figure 2.1). The site is managed by the Forestry Administration (FA),
with technical and financial assistance provided by the Wildlife Conservation Society
(WCS) as part of a long-term conservation programme. SPF is located in part of a
large former logging concession held by the Malaysian company, Samling
International Ltd. Commercial logging operations in this concession were halted in
1999, due to increasing royalties for felled timber, and a nationwide moratorium on
logging was introduced in 2002. Since 2002, the site has been managed for
biodiversity conservation and local livelihoods enhancement. In September 2009, the
area was gazetted as a formal Protection Forest by Prime Ministerial Subdecree #143,
which separated the site into a core protection zone (1885 km2) and buffer zone (1042
km2).
Figure 2.1: Location of Seima Protection Forest in Cambodia (from Evans et al., 2012a).
17
Biophysical characteristics
The site contains a complex mosaic of habitats types, ranging from sparse deciduous
dipterocarp forest in the lowland northwest to dense tropical evergreen forest in the
eastern region along the Vietnamese border (Walston et al., 2001, Evans et al., 2003).
The terrain is largely hilly and elevations range from 100m in flatter lowland areas up
to 700m on the Sen Monorom Plateau. The site is of significant importance for
biodiversity conservation, containing at least 41 animal species listed as globally
threatened (Evans et al., 2012a).
Human geography
History
Cambodia is home to several minority indigenous peoples, the majority of whom live
in remote forested upland areas (ADB, 2002), often in and around gazetted protected
areas such as the study site. Although little is known about the remote forested areas
of Cambodia prior to French colonial rule, it is thought that the study area was
sparsely settled by several indigenous groups, who maintained trading links with
lowland areas (McAndrew et al., 2003). During French rule, aside from a small
number of rubber plantations, the area remained largely autonomous (Evans et al.,
2003). However, following independence from France in 1954, Cambodia’s minority
indigenous peoples came under increasing pressure from successive regimes to
integrate into Khmer society, culminating in mass forced resettlements during the
Khmer Rouge period (1975-79). At this time, villages within SPF were cleared and
the majority of people moved to Khao Gniek District in the lowland north of
Mondulkiri Province (McAndrew et al., 2003). While some households began to
return to their villages after the fall of the Khmer Rouge, the majority of returnees did
not do so until the 1980s and 1990s.
The study site now supports a rapidly increasing human population. In 2008,
population density had reached 5.4 people/km2, with the significant majority of
households belonging to the indigenous Bunong people (the original inhabitants prior
to the Khmer Rouge period; Pollard & Evans, 2008). The remaining population is
comprised chiefly of Khmer immigrants, who have moved to the area from
neighbouring lowland provinces in search of land, or who belong to former Khmer
18
Rouge cadres that have been demobilised in the area. In addition to migrants who
have settled permanently in the project area, there is also a significant transitory
migrant labour force that travels to the uplands from neighbouring provinces as a
means of extending the agricultural season. These temporary workers rarely feature in
official statistics but are thought to be an important component in the pattern of
agricultural expansion seen in the project area because of the extra labour they
provide.
Bunong culture and institutions
A typical Bunong household is made up of a married couple, their children (including
those who have married but not yet moved out) and any surviving parents who are too
frail to live alone. Household members jointly farm their land and mostly share food
and income. Many of the households in a typical settlement are closely related by
blood or marriage and there are strong traditions of sharing and interest-free loans
between family and close neighbours, providing a key livelihood coping strategy
(Evans et al., 2003). Most households have two residences, a permanent house in the
main settlement and a smaller camp close to their fields, which may be some distance
away.
The Bunong have a deep connection with the natural environment. As with most of
the minority indigenous peoples in Cambodia, the Bunong have a strong animalistic
belief system (Kuoy, 2013), which is linked to many rituals and ceremonies
throughout the year in worship of the spirits that they believe inhabit the hills, forest
and streams around each settlement (Evans et al., 2003). These beliefs are
incorporated into customary laws, which prohibit clearance of forest in the immediate
vicinity of settlements, in burial forests or in so-called spirit forests (Fox et al., 2008).
Traditionally, the governance structure was village-based, with no higher-level
institutions (Scott, 2009). Today, traditional village level institutions survive
alongside state administrative roles (e.g. village chief), protected area authorities (e.g.
FA, police and military police) and NGOs (e.g. WCS and local development partners),
such that institutional controls on behaviour are both formal and informal. Village
elders (purahn ban) play important roles resolving disputes, presiding over
ceremonies and advising other villagers about the harvest of natural resources or
clearance of forest (UNDP, 2010). The spokesman for the village elders (antreahn)
19
also performs a key mediation role, brokering resolutions to disputes and awarding
sanctions for those considered to have broken rules. Sanctions typically take the form
of gifts, such as chickens or rice wine, to aggrieved parties or to the whole village if
the offence is judged to be serious. In several villages, these institutions have merged
such that the traditional role of mediator often falls to the village chief, a political
appointment. Internally within the village, new institutions are emerging through the
creation of resource management committees associated with various NGO
interventions. While these committees are ostensibly based on traditional institutional
structures, with village elders making up the majority of committee members, the
roles and responsibilities of the committees are more formalised and larger in scope
than would traditionally have been the case.
Livelihoods
Although livelihood strategies throughout the project area depend on the terrain, the
majority of indigenous households would traditionally have cultivated hill rice and a
diverse range of vegetable and fruit crops to meet their subsistence needs,
supplemented by the consumption and sale of forest products (most notably liquid
resin collected from tapped dipterocarp trees). In a survey of resin-tappers, Evans et al.
(2003) found that the sale of liquid resin on average contributed $338 to total
household incomes annually. In upland areas, rice is traditionally grown in rain-fed
sloped fields, which are cultivated for a period of two to three years before being
allowed to return to fallow. This type of farming is carried out in a rotational system
with new fields cleared on a regular basis from land previously set to fallow. However,
these rotational periods were severely disrupted by the forced evictions carried out by
the Khmer Rouge, with the result that many areas that would previously have been
part of the rotational cycle now hold 40 years or more of forest growth. In lowland
parts of the project area, there is some rotational farming but the majority of
households practice more productive, but sedentary, paddy rice cultivation. As
traditional livelihood systems relied heavily on small-scale swidden agriculture, only
a modest proportion of communal lands were cultivated at any one time. While
swidden land use has often been considered to have a negative impact on forest
ecosystems and subjected to efforts to ‘stabilise’ agricultural production (e.g. Lao
land reforms; Ducourtieux et al., 2005; Lestrelin, 2010), there is evidence to suggest
that such systems can be sustainable (Fox et al., 2000).
20
Although the majority of indigenous households in project villages maintain
traditional livelihood strategies, there has been an increasing trend towards the
adoption of commercial agriculture observed throughout the uplands region (Fox et
al., 2008; Milne, 2013). This is bringing more and more families into the market
economy, although it should be noted that the Bunong have a long history of trading
(McAndrew, 2009). The introduction of cashew, and more recently cassava, is
thought to have contributed to significant expansion in the area of land farmed per
household (T. Evans, pers. comm.). This expansion is also thought to have affected
fallow cycles, with many plots now being planted with cashew rather than being
allowed to return to fallow.
Site management
The majority of Cambodia’s protected area network was created during the 1990s,
and is managed by the Ministry of Environment (MoE). However, little was
understood about the areas gazetted due to on-going conflict during this period. As
such, PAs created at this time were little more than paper parks (Wilkie et al., 2001),
and many continue to be so. From 2002, a series of Protected Forests have been
created by Prime Ministerial sub-decree and are managed by the FA. Consequently,
the Cambodian protected area network is made up of sites managed either by the MoE
or FA. Most of theses site are located in remote forested areas, with lower population
densities and higher poverty levels than more accessible parts of the country (World
Bank, 2009). Under the 2005 Forestry Law and 2008 Protected area Law, livestock
grazing and the extraction of “traditional subsistence” levels of timber and NTFPs by
local communities are permitted. Commercial logging and clearance of forest are,
however, illegal.
The main threats to wildlife at the study site come from the illegal clearance of forest
and both commercial and subsistence hunting (Walston et al., 2001, Evans et al., 2003,
Evans et al., 2012a). Until 2012, the majority of forest losses were thought to be
caused by small-scale agricultural expansion and clearance by immigrant households.
In addition to these small-scale land grabs, there has been a significant increase in the
area of forest cleared by economic land concessions (ELCs). The granting of these
large scale, agro-industrial plantations has increased dramatically across the country,
with ELCs covering 381,121 ha granted in 2012 alone (of which over 70% were
21
within existing protected areas; ADHOC, 2013). In response to growing unrest, a
moratorium on the granting of new ELCs was introduced in 2012, although
applications that are claimed to have been “in the pipeline” continue to be approved
(ADHOC, 2013). In addition to land grabs, another threat to wildlife and local
livelihoods is illegal logging. Loggers commonly target high value timber species,
including those tapped by local villagers for the collection of liquid resin, and hunt
while camped in the forest. Recent discussions with villagers throughout the protected
area have highlighted the loss of resin trees as a growing concern, particularly in areas
located close to those granted as ELCs (unpublished data). Conservation activities to
mitigate these threats have focused on law enforcement, community engagement,
biological monitoring and participatory land use planning. Consequently, there are
many more controls on resource use than in the past, with timber, land and certain
wildlife species subject to legal protection. However, the collection of liquid resin is
permitted throughout the site and the trees from which the resin is collected are
protected under law. Sanctions for those individuals caught extracting resources
illegally are minimal and typically comprise the confiscation of tools and whatever
harvest the individuals have been caught with.
Proposals for carbon financing of forest conservation in the area are well advanced,
and are being led by WCS at the request of the Cambodian government. These
proposals focus on the core protection zone, with participating villages located within
or in close proximity to this zone (Figure 2.2). These villages are divided into 17 key
villages (those with farmland or residential land in the project area) and three other
user villages (those documented to have regular, significant forest use in the project
area but no agricultural or residential land inside). In the key villages, the whole
village is involved in most aspects of the project, as most or all families are natural
resource users; in the other user villages project activities are focused only on those
families identified as being regular users of the project area (which in certain cases is
only a small minority of all families). The majority of villages within the site are
comprised of a number of discrete settlements, which can often be located several
kilometres from each other but still belong to the same administrative village unit. In
many cases similar conditions are found in each of the settlements but in one or two
cases livelihood opportunities may be significantly different (for instance if one
settlement is more inaccessible than others).
22
Figure 2.2: Landscape showing the location of the 17 key villages and 3 other user villages participating with the project and identified for inclusion in the social
monitoring framework (from WCS, 2013).
23
Under the Seima Protection Forest REDD+ Demonstration Project, WCS is providing
technical assistance to the FA to develop a site-based carbon offset project under the
REDD+ framework. Credits will be generated in the core protection area of SPF,
which is a designated national REDD+ demonstration site (FA, 2010). The project is
currently undergoing validation against the two leading voluntary market standards,
the Verified Carbon Standard (VCS), which focuses on quantifying emission
reductions, and the Climate, Community and Biodiversity (CCB) Standard, which
focuses on assuring social and environmental co-benefits. Conservative projections
suggest the site can generate emissions reductions of 300,000 tCO2e per year beyond
baseline levels (WCS, 2013). Revenues from the sale of credits generated by the
REDD+ project are expected to expand opportunities for conservation activities,
including greater emphasis on incentive based approaches (Milne et al., 2012).
24
Chapter 3
3. Assessment of progress towards social targets of the Seima
Protection Forest conservation programme
Introduction
There is increased awareness of the need for more effective monitoring and evaluation
of conservation outcomes and impacts. Enhanced monitoring would enable
conservation practitioners to assess the effectiveness of conservation policies, add
transparency and accountability to decision-making, detect potential issues and
identify possible remedial actions (Margoulis et al., 1998; Ferraro & Pattanayak,
2006; Stem et al., 2005). Where commitments have been made to minimise or
eliminate negative impacts on local people or to deliver positive benefits, effective
social monitoring and assessment of impacts is essential (Brockington & SchmidtSoltau, 2004). To be effective, social impact assessments must be able to differentiate
between different groups and measure changes in well-being across groups,
particularly those considered to be most vulnerable to environmental change (Adger,
2006; Daw et al., 2011). Such assessments can also bring practical benefits to the
implementation of conservation policies, particularly in early stages, by improving
understanding of the social context in which conservation operates and identifying
existing social and economic trends under current conditions (Brechin et al., 2002).
One of the key challenges surrounding any impact assessment is the concept of
attribution, defined here as the ability to ascribe particular observed social changes,
such as household well-being, directly to project activities and not other drivers of
change (Richards & Panfil, 2011). As conservation projects typically operate within
highly complex and dynamic social environments (Berkes, 2004), the concept of
attribution becomes particularly important. Without it, social changes (both positive
and negative) may be attributed to project activities, even if these changes would have
occurred if no project activities had taken place at all. Such issues can be addressed by
using quasi-experimental matched methodologies, in which control areas are
identified that resemble the project site but where no project activities are conducted
(e.g. Naughton-Treves et al., 2011; Clements & Milner-Gulland, in press; Clements et
25
al., in press). However, these approaches are costly and it can be difficult to identify
conditions that closely match project sites or that are not subject to alternative
interventions. Matched approaches have also been subject to criticism regarding the
ethics of repeatedly surveying people who are not benefiting from project activities
(Richards & Panfil, 2011). In development economics, where matching techniques are
also applied, there are concerns that matching compares poorly with the results of
randomised controlled trials (Duflo et al., 2008). An alternative approach to impact
assessment, which is recommended for use in guidance issued by the Climate,
Community and Biodiversity Association (CCBA), utilises the theory of change
(ToC; Richards & Panfil, 2011). This approach is claimed to overcome the issue of
attribution by testing a priori hypotheses regarding cause-and-effect relationships
between project activities and social outcomes, and relies upon the careful selection of
monitoring indicators that reflect attributable social changes. This has the advantage
over quasi-experimental approaches that there are no requirements to identify control
areas or survey households not directly affected by the project, but attribution is less
robust in the absence of a counterfactual.
Although the framework of reduced emissions from deforestation and forest
degradation (REDD+) brings little new to the table in terms of how conservation is
implemented on the ground (Blom et al., 2010; Clements, 2010), its emergence has
been greeted enthusiastically by some conservation organisations and national
governments, who perceive its potential to secure funding for the long-term protection
of forested areas. Whether or not this potential is realised remains to be seen. Yet due
to its regulated nature and significant investment in readiness activities, REDD+ has
been well placed to incorporate and develop requirements for the assessment of the
social impacts associated with its implementation. At the national scale, a set of seven
safeguards for REDD+ implementation was agreed at the Conference of the Parties
for the United Nations Framework Convention on Climate Change (UNFCCC) in
Cancun in 2010. Included in these safeguards, was the commitment to incentivise the
“protection and conservation of natural forests and their ecosystem services, and to
enhance other social and environmental benefits” (UNFCCC, 2010). This was
supported a year later by Decision 12/CP.17 of the UNFCCC Durban Outcome, in
which party states agreed to implement a national level safeguards information system
to systematically monitor how REDD+ safeguards are being addressed (UNFCCC,
26
2011), and reiterated by the Warsaw Framework for REDD+ in 2013 (UNFCCC,
2013). These developments have been matched at the project level, where voluntary
projects certified as meeting the Climate, Community and Biodiversity (CCB)
standards, must demonstrate that there has been a net positive impact on local
communities as a result of project activities (CCBA, 2013). As such, REDD+
activities implemented at both national and project scales should benefit local people
and include a component to assess the social impacts associated with those activities.
In order to satisfy the requirements for social impact assessment for projects applying
for certification under the CCB standards, PA authorities at SPF designed a social
monitoring framework for the study site (WCS, 2013). This framework is based on
the ToC-based approach to social impact assessment of REDD+ projects
recommended by the CCBA (Richards & Panfil, 2011). Under this methodology,
“causal chains” are used to conceptualise linkages between project goals and activities.
To do this, project proponents made use of a conceptual model previously created in
2006, and updated in 2011 as part of the development of the REDD+ project at the
site, which linked project targets to conservation activities and to direct and indirect
threats (Figure 3.1; WCS, 2013). Indicators were selected to measure progress
towards each target.
In this chapter, I investigate social trends in the 20 villages participating in the
REDD+ demonstration project at Seima Protection Forest for the period between
2007 and 2012, utilising the indicators selected for the social monitoring framework
described above. The aim of this analysis is threefold:
1. to provide a baseline picture of trends in site demography, household
economic well-being and resource security to inform the analyses presented in
Chapters 4, 5 and 6.
2. to evaluate the social impacts of activities implemented at the study site in
advance of REDD-linked social programmes in relation to two key stated
targets of conservation management of the site:

increase security and productivity of natural resources to support local
livelihoods

ensure sufficient farmland to support the livelihoods of current
residents
27
GOAL
A well-managed forest landscape that supports increasing wildlife populations and improving livelihoods for the people who currently live there
TARGETS
DIRECT
THREATS
Maintain the variety, integrity,
and extent of all forest types
Increase populations of wildlife of
conservation concern
Clearance for land
concessions and
other projects
Increase security and productivity of natural
resources to support local livelihoods
Forest clearance/grabbing by individuals
Over-fishing, over-hunting of wildlife
illegal logging and overexploitation of NTFPs
Sufficient farmland to support the
livelihoods of current residents
Land alienation and
legal conflict
Limited land
productivity
[Impacts not
yet known]
INDIRECT
THREATS
DIRECT
INTERVENTIONS
SUPPORTING
INTERVENTIONS
Undefined
borders and
regulations for
the SPF
1. Develop the key legal
and planning documents
needed to manage SPF
Weak traditional
institutions and
lack of voice
3. Establish sustainable
community use of land and
natural resources; adapt to
climate change
2. Reduce forest crime
through direct law
enforcement
5. Effective monitoring
Population
growth, inmigration, better
access
6. Effective administration
Scarcity of sustainable
dev. livelihood
opportunities, on and off
farm
4. Support alternative
livelihoods that reduce
pressure on forest and NR;
adapt to climate change
7. Sustainable finance
Figure 3.1: Conceptual model developed by PA authorities for Seima Protection Forest REDD+ Demonstration Project (from WCS, 2013)
28
Future
climate
change?
3. to investigate the appropriateness of the indicators selected for these two
targets as part of the monitoring framework for social impact assessment at the
site.
For this analysis, I designed a household survey, which was undertaken in 2012, to
supplement existing monitoring activities at the study site and to build upon the
results of a complementary survey undertaken across the same landscape in 2007.
Where data were previously lacking, the 2012 survey creates a baseline for the
indicators selected by project proponents for the assessment of social co-benefits and
impacts expected to arise from the REDD+ project under development at the site. For
data collected in both 2007 and 2012, progress towards project targets is evaluated by
assessing the trends in selected indicators. As such, the 2012 survey forms a key
component of the social monitoring system developed by PA authorities for the
REDD+ project. In addition to the analysis of social trends and impacts of
conservation activities at the study site, I present a discussion of lessons learned from
the implementation of this survey and associated implications for future social impact
assessment at the site.
Methods
Indicator selection
For each indicator selected to measure progress towards the two project targets
concerning local livelihoods, I identified the expected trend that would indicate that
progress had been made towards the respective project target (Table 3.1). In all cases,
the business as usual trend (i.e. in the absence of project activities) was thought to be
negative, and so a positive trend was not necessarily required to indicate a positive
impact, as a stable trend could indicate that the project was having at least some
positive effect. In these cases, however, an observed negative trend would be
inconclusive in light of a lack of a counterfactual.
29
Table 3.1: Selected indicators for the two social management targets identified for Seima
Protection Forest. The expected trend that would indicate progress towards project targets is
given for each indicator.
Project target
Indicator
1. increase security
and productivity of
natural resources to
support local
livelihoods
1. resin tree ownership:
 1a. no. of households
 1b. no. of trees per
household
With-project expected trend
(2007-2012)
increased or stable security,
abundance and productivity of
harvested natural resources
1c. reported number of
meat meals in a week
1d. reported household
resource security
2. ensure sufficient
farmland to support
the livelihoods of
current residents
2. land ownership
measures:
 2a. average holdings
 2b. % landless
increased average area of
household land holdings, stable
or reduced proportion of
landless households
Target 1: increase security and productivity of natural resources to support local
livelihoods
Security of natural resources can be understood in a number of different ways, which
may reflect aspects of tenure (i.e. the legal rights of resource users to access, use or
claim ownership of resources) or the sustainability of resource use. In the context of
the project target evaluated in this chapter, resource security is taken to mean the
long-term supply of the forest-derived natural resources upon which local livelihoods
depend. The project target is, therefore, interpreted as a stated intention to manage the
study site in such a way as to ensure the long-term protection and sustainable supply
of key natural resources within SPF. This is a difficult quantity to measure, as changes
in the use of different forest products may be caused by a number of factors, such as
varying demand or changes in stock levels. For this reason, three separate indicators
were selected to measure progress towards the target in addition to a poverty score
derived for each household. These included two variables focussed on household
access or use of resources (household resin tree ownership and the number of wild
meat meals eaten by a household in a week) and a third variable looking at reported
perceptions of household resource security as defined above.
30
Resin tree ownership:
As liquid resin has previously been shown to provide a significant proportion of
household income at the project site (Evans et al., 2003), the trees from which resin is
collected are considered to be an important household resource. These trees are
subject to customary tenure rules, which give ownership rights to individual
households. As such, tree ownership is an important indicator of the security of
natural resources that support local livelihoods. The total number of trees that can be
tapped is finite and considered to be saturated (i.e. all trees mature enough to be
tapped are already owned by a household; Clements, 2012). The transfer of ownership
is largely through inheritance. When children reach marrying age and form new
families, they are typically given a small number of trees by their parents. Few trees
are reported to be sold. The biggest threat to resin trees comes from illegal logging,
particularly in areas close to economic land concessions (H. Travers, pers. obs.).
Although local communities respect customary ownership of resin trees, illegal
logging teams target resin tree species because of the high value of the dipterocarp
timber. Consequently, the loss of trees to illegal loggers has the potential to reduce
total resin tree stocks and, hence, undermine local livelihoods. Stability in the number
of households that own trees and the number of trees owned by resin tapping
households would indicate that resin trees are being successfully protected. As resin
tree species are some of the most valuable timber species in the project area, resin tree
numbers also provide an indication of the success of wider forest protection.
Meat meals:
Wild protein sources are important for rural households, particularly the poor (MilnerGulland & Bennett, 2003). As such, the number of meals eaten that contain wild meat
(including fish) could be an indicator of natural resource security. Stable consumption
would indicate that wild meat resources are sufficiently stocked to be able to continue
to support extraction, although this would not necessarily indicate that extraction
levels were sustainable into the longer term. However, as households within the
project area become better off or more farming-dependent, some households may
switch to alternative meat sources, such as farmed livestock or imported fish, as has
been noted in other contexts (Brashares et al., 2004; Wilkie et al., 2005). A decline in
the average number of wild meat meals eaten under such circumstances would not
necessarily indicate a decline in wild stocks. However, a decline in the number of
31
wild meat meals concurrent with a decline in the total number of meat meals eaten
(from both wild and farmed sources) could be expected to be indicative of a reduction
in wild stock levels. Hence this ratio was chosen as the indicator.
Perceived resource security:
The final indicator selected to measure household resource security was simply the
reported perceptions of those households interviewed for the 2012 livelihood survey.
Each respondent was asked whether they felt secure about their access to natural
resources, including agricultural land, resin trees and other forest resources, and the
reasons why they felt this way.
Target 2: sufficient farmland to support the livelihoods of current residents
The second target identified in the conceptual model for the project is to ensure that
sufficient farmland is available to support current livelihoods. Limited land
productivity is a direct threat to sustainable livelihoods in the project area, as low
yields and extensive land use practices increase the demand for smallholder farmers to
clear more forest in order to meet their production needs. Two indicators were
selected by WCS to monitor this target: average household land holdings and the
proportion of households without any land at all.
Household land holdings:
Household land use patterns have changed dramatically in the period between the two
surveys in 2007 and 2012. In part this is due to the emergence of cassava, which was
introduced in 2005/6 by Vietnamese traders. This crop has changed the prevailing
agricultural system practiced in the project area, with a wave of cassava adoption
spreading throughout SPF as access to villages improves and the prices offered by
traders increase. The emergence of cassava, and to a lesser degree other cash crops,
has increased the incentive for farmers to claim extra land by clearing more forest. It
is expected that increased land tenure security amongst many of the project villages as
a result of project activities will encourage farming households to make the most of
this opportunity, whilst restricting encroachment into conserved areas (see Chapter 4).
Of the 20 project villages, 15 villages have received or are in the process of applying
for communal land tenure in recognition of customary land rights (A. Diment, pers.
comm.). Land designated to communal title is managed by a village committee, which
32
then grants plots to individual households. For many households, this process is
expected to give them access to greater areas of agricultural land and, hence, the
average area of household land holdings is an important indicator of the success of
this programme. However, unrestricted expansion of agricultural lands remains a
threat. As such, it is important also to monitor the number of households that hold 5
ha or more land. The threshold of 5 ha was selected, as this is the area that households
are limited to by community regulations governing communally titled areas or legally
in the case of individual land titles.
Land alienation:
As commercialisation of agriculture spreads throughout the project area, household
livelihoods are increasingly dependent upon access to land. As such, households are
becoming progressively more vulnerable to land alienation. At the same time,
pressure from immigrant households and influential external interests is serving to
increase the price of land and draw in additional land speculation. More widely, land
disputes between local communities and large companies have increased dramatically
throughout Cambodia (ADHOC, 2013). Whilst SPF has largely been unaffected by
the impacts of the widespread granting of ELCs, three rubber concessions have
recently been granted in the buffer zone, illustrating the increasing threat of
dispossession and land alienation to local livelihoods. Minority indigenous
communities, such as those found in the project area, are particularly vulnerable to
dispossession as a result of debt, misinformation, weak political voice or extortion. In
order for the project to achieve its target of sufficient land for local communities, it
will be necessary to simultaneously protect communities from the threat of land
alienation. As such, the extent of landlessness within the project area is an important
indicator against which to measure success.
Vulnerable groups
One of the key requirements of any social impact assessment is to be able to detect
project impacts on different socio-economic groups within target communities,
particularly those considered most vulnerable to change or adverse impacts. To this
end, I identified several potential vulnerable groups (Table 3.2).
33
Table 3.2: Hypothesised vulnerable and special interest groups identified for the impact
assessment. These groups are not exclusive and so it was possible for households to be classified
as belonging to more than one group.
Potential vulnerable Reason for testing
households
widow-headed
widow-headed households are likely to be more vulnerable
to shocks and have fewer livelihood opportunities.
dependent
households that consist entirely of an elderly couple (and
possibly grandchildren) may be more vulnerable to shocks.
These households are also likely to have fewer physical
resources as these are often given to their children upon
marriage.
landless
landless households lack the means to grow their own food
and may be dependent on daily wage labour opportunities.
Landlessness can also be a sign that the household has fallen
into debt and has sold their land as a consequence. Not all
landless households are poor, as some households may
consciously choose to focus on other livelihood activities.
non timber forest
product (NTFP)
collectors
NTFP collectors have been classified here as non-timber,
non-liquid resin collectors. Commonly collected NTFPs
include rattan, bamboo, honey and wild fruit and vegetables.
Although there is a degree of overlap between the two
groups, many NTFP collectors do not have resin trees to
support them. NTFPs are thought to offer a safety net to
households that experience hard times, and those households
that collect them may be particularly vulnerable to losses in
forest resources.
labour sellers
households that are dependent on daily wage opportunities,
which are low in some villages, are more likely to have poor
access to important resources, such as land and resin trees.
indigenous
as the majority of households in the project area are
indigenous, there is little evidence to suggest that they are
more vulnerable than any other group. It is possible that this
may change in the future, however, as there have been
increasing numbers of migrant households and indigenous
households can be vulnerable to land alienation.
resin tappers
resin tree owners are not necessarily a vulnerable group,
having access to a resource that is less subject to fluctuating
yields than some crops. Tappers with few trees are more
vulnerable to losses, however, and this group is of special
interest to the project as it seeks to protect forest resources
with the project area.
Once vulnerable groups were selected, trends for these groups could be investigated
in addition to those for the overall population. In some instances, certain groups were
34
identified for monitoring despite not proving vulnerable overall because of their
special interest to the project (e.g. resin tappers, indigenous households).
Household livelihood survey
Twenty target villages were identified through initial consultations between PA
authorities and potential forest users in the REDD+ project area. A quantitative
household livelihood survey was conducted in all of these villages between April and
July 2012. This survey was designed to complement a livelihood survey conducted in
2007 to create a social baseline against which impacts of the REDD+ project could be
assessed. As such, the survey was designed for repeated iterations over the period of
the REDD+ project, which influenced some aspects of the design. The survey
interviews were conducted by a Cambodian organisation, the Centre for Development
Orientated Research in Agriculture and Livelihood Systems (CENTDOR), which
specialises in independent monitoring and evaluation of rural development projects.
The survey was carried out in this way in order to minimise the risk that interview
respondents tailored their responses to what they thought PA authorities wanted to
hear.
In all bar three villages, at least 30 households were randomly selected for inclusion
in the survey, with 622 households interviewed overall. In the remaining three
villages, there were insufficient households in the village at the time of the survey to
be able to sample at least 30, and so all available households were sampled. For
villages that are separated into discrete settlements, the sample was proportionally
stratified by settlement. For example, if a village comprised two settlements of 200
and 100 households, the team would survey 20 households in the larger settlement
and 10 in the smaller. Within each settlement, households were randomly selected
from a list of households kept by the village chief. Where possible, interviews were
conducted with those individuals identified as the household head. In cases where the
household head was unavailable for extended periods, their spouse was interviewed
instead. Interviewers followed a structured questionnaire with a final qualitative
section focusing on perceptions of livelihood changes. All interviews lasted for
approximately one hour, with respondents compensated for their time.
The 2007 survey was conducted by the monitoring and evaluation unit of Cambodian
Centre for Development and Study in Agriculture (CEDAC), which was subsequently
35
established as CENTDOR in 2008. For this survey, 400 households from 14 villages
were interviewed, with sampling approximately proportionally stratified by village,
such that the majority of interviews were conducted in more populous villages. As in
2012, households were randomly sampled in each village and interviews were
conducted with the household head.
The basic necessity survey
Although the 2012 survey was designed to produce compatible results with those
produced by the 2007 survey, there were methodological differences between the two
surveys, so that improvements could be made to the protocol used and to the content
of the questionnaire. The most significant of these differences was in the handling of
household economic well-being. The 2007 survey focused heavily on the collection of
income data but as livelihood strategies within project villages are vulnerable to
stochastic events, such as poor weather or flooding, income data can be highly
variable both spatially and temporally. This can make comparisons between different
parts of the landscape or different years difficult. For example, the rice harvest of
2006 was particularly poor and many households who were interviewed during the
2007 survey had been badly affected. Conversely, the rice harvest of 2012 was
generally good. Hence, any comparison of incomes between the two years would be
biased and likely to suggest a greater improvement in economic well-being than was
actually the case.
In order to overcome this issue, an alternative metric of household economic wellbeing was selected: the household poverty score derived from following the basic
necessity survey method (Davies, 1997). This is a survey instrument that provides a
participatory poverty index of sampled households, which is derived from ownership
and access data for a list of household assets and services. For the 2012 survey, a list
of 40 different assets and services was compiled at a workshop held at the FA’s Seima
headquarters in April 2012 (see CENTDOR (2012) for a more detailed description of
the workshop). This workshop was attended by 24 participants from villages located
in the three different livelihood zones, who were selected to represent different socioeconomic groups with each village by project staff with a good knowledge of each
community. Following the adapted definition used by Clements et al. (in press), basic
36
necessities were defined as the “minimum requirement for living that all households
of the community should have and no-one should not have.”
Each household sampled was asked whether they had access to or owned each item on
the list. For each item the respondent was also asked whether they believed that item
to conform to the definition of a basic necessity. Each household’s poverty score was
calculated by adding the weighting for each item (calculated from the proportion of
respondents who agreed that an item was a basic necessity) that that household had
access to or owned, provided that at least 50% of respondents thought that those items
were basic necessities. For example, if a household owned four items, which were
thought to be basic necessities by 41%, 22%, 86% and 79% of respondents
respectively, their poverty score would be 0.86 + 0.79 = 1.65. These scores were then
converted to a proportion of the total possible score. In 2012, 32 of the original 40 list
items were thought to be basic necessities by over 50% of households sampled. Items
that less than 50% of people would think were basic necessities were included in the
list to make respondents think carefully about whether each item conformed to the
definition and to future proof it against potential increases in wealth.
Comparison of 2007 and 2012 household poverty scores
A basic necessity survey component was not included as part of the landscape-wide
livelihood survey in 2007, but data were collected on 11 assets and services that
appeared as items in the 2012 list. Two methods were tested to derive an adjusted
poverty score for 2007, using ownership data for just these 11 items.
The first method simply found a reduced poverty score for the 11 items for which data
was collected in 2007 by summing the 2012 weightings for each item owned. This
gave a reduced score based on 2012 perceptions of whether or not those items met the
definition of a basic necessity. In the second method, a new score for households in
2012 was calculated using the 21 items for which no data was collected in 2007. This
adjusted score was then regressed against dummy ownership variables for the 11
items using the linear mixed modelling method described below, which provided
linear coefficients for each item plus a best linear unbiased predictor for each village.
A predicted poverty score for the 21 items not covered by the 2007 survey was then
calculated from these coefficients and added to the score for the 11 items included in
37
both surveys. This gave an estimate for both time periods of the poverty score derived
for all 32 items, as opposed to the score for only the 11 items common to both surveys
given by the first method.
Before comparisons could be made between the two time periods, it was necessary to
test the extent to which the adjusted poverty scores calculated using the two methods
were representative of the score calculated for the full 32 items used to calculate the
overall poverty score for each household in 2012. Pearson's product-moment
correlations comparing the adjusted scores with the full poverty score were found for
households sampled in 2012. These showed that the reduced score derived using the
first method provided a closer correlation (correlation coefficient = 0.90) with the
original full score and so this method was preferred as the basis for making
comparisons between the two time periods.
In order to compare the reduced poverty scores from the two surveys in greater detail,
a linear mixed model was constructed following the approach described below, with
reduced poverty score as the response variable. The final model included some
explanatory variables that were different to those used to investigate the overall
poverty score in 2012. Two changes were necessary because the survey data collected
in 2007 did not allow for the calculation of household adult male equivalence and no
data were collected regarding labour constraints. Consequently, total household size
was used as a substitute for adult male equivalence and whether a household was
labour constrained was omitted from the model. A third change was made to
substitute a dummy variable for landlessness for the reported area of land claimed per
household, as landless households are one of the vulnerable groups identified above.
Finally, a survey year dummy variable was added to test for changes in poverty score
between the two surveys and an interaction term between this dummy variable and the
distance to the nearest market was also included.
Data analysis and statistical modelling
One of the main issues that had to be addressed before making comparisons between
the results of the 2007 and 2012 surveys was the difference in sampling density
between the two surveys and the difference in the villages included in each survey. In
2007, sampling in each village was done on an approximately proportional basis, with
the consequence that very few interviews were conducted in some of the smaller
38
villages. In 2012, the sampling design was modified to make it easier to monitor
variations in indicator trends between villages over time. Consequently, different
sampling densities were used for the two surveys. For each time period, weighted
means and standard errors were calculated for each indicator using sampling weights
for each village equal to the number of households in a village at the time of the
survey divided by the number of households sampled during that survey. As the
number of villages included in the survey was increased in 2012 to ensure that all 20
project villages were monitored, comparisons are made solely on the basis of those
villages that were sampled in both surveys but excluding Kmom and Sre Andaol, as
land use data for these two villages were considered unreliable for the 2007 survey.
Due to the sampling design of the two surveys and the overall aim of the monitoring
programme, analyses of longitudinal trends were based on comparing the results of
cross-sectional analyses of each time period (rather than a panel analysis), as it was
felt that this approach offered multiple advantages over the length of the REDD+
project. Cross-sectional sampling does not carry the risk of participant attrition to
which repeat surveys are vulnerable, nor does it put so much of a burden on
respondent households. Furthermore, newly arrived or formed households would
continually need to be added to a repeated survey. Finally, it can be very difficult to
match households between surveys, particularly if the household head dies, members
move away or the children marry and form new households.
Statistical modelling was conducted to give a greater understanding of indicator
trends and the socio-economic groups associated with different activities. Continuous
variables were analysed using linear mixed models (LMMs) and binomial or
Bernoulli variables were analysed using generalised linear mixed models (GLMMs).
For each analysis, a number of household demographic and livelihood variables were
investigated as explanatory variables (see Appendix A; Table A.2 for a description of
the household variables investigated). For analyses in which the data from all sampled
households were included, the area of land reported by each household was
transformed by taking the natural logarithm of reported area plus the square of the
first quantile divided by the third quantile. This constant was added to account for
zero values.
39
Backwards stepwise model selection was carried out on the basis Akaike information
criterion values (AIC; Akaike, 1974). If candidate models had a ΔAIC value of less
than two then the most parsimonious model was selected, otherwise the model with
the lowest AIC was selected (Burnham & Andersen, 2002). Marginal and conditional
R2 statistics were calculated for LMMs following the method proposed by Nakagawa
& Schielzeth (2013). These can be interpreted as the variance explained by the fixed
effects and whole model respectively. Selected models were checked for residual
normality, heteroskedasticity and correlations between fixed effects and the residuals.
Over-dispersion in binomial logistic regression models was checked by comparing the
sum of squared Pearson residuals with the approximate residual degrees of freedom.
No issues were found.
Results
Population and immigration
In 2006, the total number of households in the project villages and settlements for
which data are available for both time periods (17 villages plus two settlements) was
1838 (Evans, 2007). In 2012, the household population for the same villages and
settlements had grown to 2768, representing an overall annual growth rate in the total
number of households of 7.1%. However, this growth was not distributed equally
between the project villages. Three villages (O Am, O Rona and Sre Preah) alone
account for 75.3% of the increase in the total number of households and if these
villages are excluded from the calculations, the resultant annual growth rate in the
total number of households is reduced significantly to 3.8%. As these three villages
are located closest to the district capital of Keo Seima District (Figure 2.2), this result
suggests that much of the overall increase in the number of households is due to
immigrant Khmer households from other provinces moving into the project area in
search of land. This is further evidenced by the resultant shift in the proportion of
indigenous households in the project area. In 2006, the percentage of indigenous
households in comparable project villages (i.e. those with sufficient data in both time
periods to be able to make a comparison) was 67.2% but this had dropped to 54.6%
by 2012, a reduction of over 10% in just 6 years. Until a recent drive to title
household claims to land in 2012/13, the vast majority of land claims from migrant
households were illegal. However, land claimed by immigrant households was
40
excluded from negotiations over indigenous land title, which reduced the legal
entitlement of the native Bunong communities to the natural resources on which their
livelihoods depend. As of yet, this wave of immigration has been restricted to the
southern lowland part of the protected area but recent improved road access to other
parts of the project area has already increased the level of immigration (H. Travers,
pers. obs.). As such, any conservation activities planned for the project site must take
account of the effect that migrating households may have.
Household poverty score
The main metric used for assessing the economic well-being of households from the
20 participant villages was the score derived from the basic necessity survey, which
ranges from 0 to 1 (with higher scores indicating better off households). The poverty
scores for households sampled in 2012 range from 0.10 to 0.82 (for weightings used
to calculate household poverty scores see Appendix A; Table A.1). This clearly
illustrates the extent of the difference in economic well-being between the poorest and
best off households in the project area. The distribution of scores is approximately
normal with a mean score of 0.44 (SE = 0.13; Figure 3.2). Once differences in the
number of households in each of the villages across the project area were accounted
for, the adjusted average household poverty score for the landscape increased to 0.47
(SE* = 0.04).
Figure 3.2: Distribution of full household poverty scores from the 2012 quantitative household
livelihood survey.
A linear mixed model was constructed in order to investigate the demographic and
livelihood factors that were most strongly correlated with household economic well-
41
being, with household poverty scores as the response variable and village as a random
effect (Table 3.3). Included in the final selected model were nine explanatory
variables: three related to household demographics and six related to livelihood
strategy or opportunities. This correlates very closely with a similar analysis of basic
necessity poverty scores for households in northern Cambodia (Clements et al., in
press).
Table 3.3: Coefficient estimates for the LMM with the full household poverty score for 2012 as
the response variable. Standard errors are given in parentheses. 𝑹𝟐𝒎 = 0.363, 𝑹𝟐𝒄 = 0.416. Resin
tree ownership at the intercept was taken to be 0-50 trees.
Variable
intercept
adult male equivalence
widow
ln(area of land + 0.3738)
owns 51-150 resin trees
owns > 150 resin trees
shop owner
labour seller
employed
service provider
labour constrained
distance to nearest market [km]
Estimate
0.487 (0.018)
+0.005 (0.002)
-0.038 (0.012)
+0.050 (0.006)
+0.031 (0.011)
+0.055 (0.017)
+0.080 (0.014)
-0.064 (0.009)
+0.034 (0.012)
+0.034 (0.015)
-0.025 (0.008)
-0.002 (0.000)
t value
27.148
2.203
-3.127
8.073
2.943
3.164
5.585
-7.355
2.892
2.268
-3.036
-5.850
The resultant linear coefficients estimated by the model provide a useful guide to
understanding which socio-economic groups are most economically vulnerable. The
results suggest that the most disadvantaged households with respect to poverty scores
were households with little or no productive land, households with few or no resin
trees, households in which the husband had died, households that were part of the
wage labour market and households that reported being labour constrained. Of the
seven hypothesised vulnerable groups presented in Table 3.2, only widow-headed,
landless and labour selling households were found to have reduced poverty scores.
Ethnicity was not a significant explanatory factor of household poverty score. This is
likely to be due in part to other variables associated with being an indigenous
household that were included in the model (e.g. resin tree ownership) but also the fact
that indigenous households are the majority ethnic group within the project area.
Interestingly, no effect was found from being an NTFP user despite the collection of
NTFPs commonly being associated with poorer households (Wunder, 2001), although
42
this is possibly because households with many resin trees (and hence wealthy) spend
longer in the forest and will collect other NTFPs at the same time.
The strongest effect on household poverty score of all those identified was for the
reported area of productive land claimed by each household. As reported areas were
log transformed and a constant added to account for skewness and the presence of
zero values respectively, interpretation of the resulting coefficient estimate is not
straight forward. Plotting the estimated relationship between household poverty score
and reported productive areas overcomes this problem and shows that the reported
area under cultivation is strongly associated with economic well-being for the range
of areas reported in the survey (Figure 3.3). It is also clear that households with little
or no land were strongly disadvantaged.
Figure 3.3: Estimated relationship between the full household poverty score and the total
reported area claimed per household. The vertical dashed line is drawn at the mean area of
productive land claimed per household.
In addition to the effects of household explanatory variables, much of the variation in
household poverty scores was accounted for by the inclusion of the village that each
household was resident in as a random effect in the model and the distance of the
village from the nearest all day market. The effect of distance to the nearest market,
which is a good proxy for the remoteness of each village, is such that for the most
remote village, Sre Khtong, household poverty scores decrease by 0.16. This is
strongly indicative of the reduced livelihood opportunities in more remote villages, as
fewer traders come to those villages and the cost of transporting goods is higher. The
residual effect associated with each village can be seen through estimating best linear
unbiased predictors (BLUPs) for the village random effect, which for the purposes of
43
this analysis can be conceptualised as the equivalent of the linear coefficients found
for the explanatory variables. Plotting the BLUPs for each village shows that, despite
quite large 95% confidence intervals, there is significant difference in the effect that
the conditions in different villages (excluding remoteness) have on household poverty
scores (Figure 3.4).
Figure 3.4: BLUPs for the village random effect. The x-axis shows the effect of living in a
particular village in terms of the difference to the full household poverty score from the intercept.
Error bars show the 95% confidence interval for each village.
Adjusted poverty scores were used to compare household economic well-being
between 2007 and 2012. For the 11 items used to create the adjusted poverty scores,
these scores ranged from 0.08 to 0.83 in 2007 and from 0 to 1 in 2012 (Figure 3.5).
Correcting for the different numbers of households in each of the villages surveyed in
both 2007 and 2012 gives average adjusted poverty scores of 0.42 (SE* = 0.07) for
2007 and 0.50 (SE* = 0.07) in 2012. Despite similar means for the two surveys, there
is a right-hand skew for households in 2012.
A linear mixed model was constructed to investigate changes in adjusted poverty
score in greater detail (Table 3.4). The effect of the survey year dummy variable
shows that there was a significant increase of 0.15 in the adjusted household poverty
score between the two surveys in 2007 and 2012. This showed that, on average,
households have access to or own more of those items classified by their peers as
44
basic necessities. As such, these results confirm that household economic well-being
has improved over the period between the two surveys. However, with the exception
of distance to the nearest market, no interaction terms were included in the model.
This suggests that there have been no significant improvements in equity, despite
overall improvements in economic well-being, as some socio-economic groups
maintain their advantage while others remain more economically vulnerable.
Figure 3.5: Comparison of the distribution of adjusted household poverty scores between the
2007 and 2012 quantitative household livelihood surveys.
Table 3.4: Coefficient estimates for the LMMs with the adjusted household poverty score as the
response variables. Standard errors are given in parentheses. 𝑹𝟐𝒎 = 0.330, 𝑹𝟐𝒄 = 0.335.
Variable
intercept
household size
widow
landless
owns 51-150 resin trees
owns > 150 resin trees
shop owner
labour seller
employed
service provider
survey year (2012)
distance to nearest market:survey year (2007)
distance to nearest market:survey year (2012)
45
Estimate
0.358 (0.023)
+0.020 (0.003)
+0.054 (0.018)
-0.064 (0.019)
+0.033 (0.015)
+0.069 (0.025)
+0.112 (0.024)
-0.116 (0.013)
+0.071 (0.021)
+0.059 (0.022)
+0.146 (0.020)
-0.001 (0.000)
-0.003 (0.000)
t value
15.542
7.145
-3.289
2.726
2.218
4.644
-8.804
3.377
7.306
-2.927
2.672
-2.218
-6.200
One exception to this can be found by comparing the effect of the interaction terms
between the distance to the nearest market and the survey year dummy variable on
household poverty scores. In 2012, the effect of living further from the nearest market
was three times stronger than for 2007. For some of the most remote villages, this
increased disadvantage due to distance from market centres is sufficient to offset the
gains suggested by the survey year. As such, households in these villages have
effectively experienced no improvement in household poverty scores. These results
suggest that, whilst overall levels of household economic well-being are improving
across the project area, development is occurring at different rates. Households in the
most remote villages have yet to experience much or any benefit, while those
households living in villages close to market centres are developing more rapidly.
Resource security
Resin tree ownership
In 2007, 216 out of 398 households (54.3%) reported owning resin trees,
approximately the same proportion as in 2012, when 301 of the 622 households
sampled (48.4%) reported owning resin trees. Applying population weightings for
each of the villages sampled in both surveys gives an adjusted proportion of 0.45
(SE* = 0.11) of households owning trees in 2007 and 0.34 (SE* = 0.15) in 2012.
Given population growth between the two surveys, this result suggests that, whilst the
proportion of households that own resin trees has decreased, the absolute number of
resin tree owners may have increased slightly (741 in 2007 to 864 in 2012). Of those
households that reported owning trees, the average number owned in villages
surveyed in both years was 93 trees in 2007 and 99 trees in 2012 respectively. This
suggests that, as both the absolute number of households with resin trees and the
average number of resin trees owned has increased between 2007 and 2012, the total
number of resin trees being tapped may also have increased. However, the differences
found between the two years are within confidence intervals and so it is not possible
to say this with any certainty. Similarly, though, there is no evidence of a significant
decrease in resin tree numbers.
In order to assess the trends in resin tree ownership in more detail, a generalised linear
mixed model, with a resin tree owner dummy variable as the response was
constructed (Table 3.5). Five explanatory variables were included in the final model:
46
household size, indigeneity, landlessness, whether the village sampled was situated in
the core zone and a dummy variable for the year of the survey. The variable with the
greatest effect was whether a household lived in the core zone, which significantly
raised the probability that households would own resin trees. Indigenous households
were also significantly more likely to own resin trees. Household size was positively
correlated with the probability of owning resin trees and landless households were
found to have a reduced probability of tree ownership, although the effects of both of
these variables were marginal. A small effect was also found for the survey year
dummy, which shows that households experienced a marginal drop in the probability
of owning resin trees between the two survey years. This drop does not necessarily
indicate a decline in the security of this important resource. The number of resin trees
suitable for tapping within a certain area is limited and it is thought that all available
trees are tapped (Evans et al., 2003). Hence, as the population grows, it is to be
expected that the proportion of households with trees might fall. Provided that the
decline in the proportion of resin tree owners was not coupled with a fall in the
number of resin trees owned per household, it is considered likely that tree numbers
were relatively stable.
Table 3.5: Coefficient estimates for the GLMM with the probability of owning resin trees. The
difference in probability was found by a reverse transformation of the logit link function used in
the GLMM for households of sample average size. Standard errors are given in parentheses.
Significance values: ** = P < 0.01; *** = P < 0.001.
Variable
intercept
household size
indigenous
landless
in core zone
survey year (2012)
Estimate
0.013 (0.065)
+0.007 (0.002)
+0.118 (0.044)
-0.023 (0.005)
+0.418 (0.260)
-0.020 (0.004)
P
***
***
***
**
**
***
To check this, a linear mixed model was constructed to estimate the number of trees
owned by each resin tree owning household, correcting for household socio-economic
factors. The only variable that was found to be correlated with the number of resin
trees owned was the adjusted household poverty score (Table 3.6), a relationship that
has already been observed (Table 3.4). The dummy survey year variable was not
included in the final selected model, suggesting that there was no significant change
in the number of resin trees owned per household between the two surveys. This
stability in the number of trees per household suggests that the observed decline in the
47
proportion of resin tree owning households is likely to be due to the increase in the
number of households in the study area rather than a decline in the absolute number of
households involved in tapping. One caveat to this result is that both the marginal and
conditional R2 statistics for the model are low, which may suggest that reporting of
resin tree numbers is unreliable.
Table 3.6: Coefficient estimates for LMMs, with the natural logarithm of trees owned by resin
tree owning households as the response variable. Standard errors are given in parentheses. 𝑹𝟐𝒎 =
0.034, 𝑹𝟐𝒄 = 0.149.
Variable
intercept
adjusted poverty score
Estimate
3.848 (0.129)
+0.080 (0.020)
t value
29.753
3.914
In 2012, respondents were also asked if they had been affected by companies or
outsiders logging their resin trees in the past five years. In total, 23 households (3.7%)
from 10 different villages responded that they had lost trees, at an average of 115 each.
It is possible that the number of trees reported as lost has been inflated in individual
cases, and in some villages it is surprising that so many trees were reported lost by
one household but no other households had seemingly been affected. While the
number of losses does not appear to be high enough to have significantly affected the
average number of trees owned by resin tapping households, the reported losses
represent a significant loss of potential income for those households affected and
indicate that some logging of resin trees is occurring within the landscape.
Wild Meat Meals
In 2007, the average number of meals that contained meat from any source was 14.2
(68% of meals), of which 4.4 meals (21% of meals) contained wild meat. In 2012, the
average number of meat meals was 11.3 (53.7% of meals), of which 3.6 meals (17%
of meals) contained wild meat. There are very strong spatial effects associated with
the number of wild meat meals consumed. As such it is difficult to explain the
differences in consumption patterns without taking account the variation between
villages. Correcting for variation between villages, by applying population weights to
calculate a weighted average, gives an adjusted population-wide average number of
meat meals 12.9 per household (of which 2.9 meals contained wild meat) in 2012.
Comparing adjusted values for households in villages sampled by both the 2007 and
2012 surveys shows that the average number of meat meals decreased from 15.8
48
meals per week in 2007 to 13.5 meals per week in 2012 and wild meat meals eaten
decreased from 5.2 to 3.1.
A generalised linear mixed model was constructed in order to understand which
socio-economic variables were correlated with the number of wild meat meals
consumed in 2012 (Table 3.7). The results of this model reveal a significant effect of
household opportunity to hunt or fish on the number of wild meat meals eaten: resin
and NTFP collectors spend more time in the forest, service providers spend longer in
the village and widows and labour constrained households have less time available to
hunt or fish than larger households or those with available labour. The effect of
household poverty score also suggests that consumption of wild meat declines with
increasing household economic well-being, although households that are in debt eat
fewer wild meat meals a week (possibly because they sell any meat that they catch).
Table 3.7: Coefficient estimates for the GLMM with the number of wild meat meals consumed
per household in one week in 2012 as the response variable. The difference in the number of wild
meals eaten was found by a reverse transformation of the logit link function used in the GLMM
for an average household (i.e. assuming sample average adjusted poverty score, household size,
education level and distance to nearest market). The difference in number of wild meat meals
consumed provides an estimate of the effect of belonging to each of the different socio-economic
groups included in the model. Standard errors are given in parentheses. Significance values: ** =
P < 0.01; *** = P < 0.001.
Variable
intercept
adjusted poverty score
household size
number of years education
resin tree owner
service provider
NTFP collector
widow
in debt
household labour constrained
feels resources secure
distance to nearest market
in core zone
Estimate
1.605 (0.530)
-0.665 (0.014)
+0.055 (0.018)
+0.093 (0.014)
+0.422 (0.113)
-0.391 (0.131)
+0.997 (0.128)
-0.311 (0.111)
-0.430 (0.070)
-0.398 (0.065)
+0.217 (0.095)
+0.040 (0.011)
+1.561 (0.861)
P
***
***
**
***
***
**
***
**
***
***
*
***
**
The estimates for village level explanatory effects (distance to nearest market and
whether the village is inside the core zone) reveal a similar picture, with households
in those villages close to all-day markets eating markedly fewer wild meat meals than
more remote villages were alternative meat sources are less available. Households
living in villages located inside the core zone were found to eat significantly more
49
meat meals per week than those living outside the core zone, suggesting that
conservation activities may be having a positive effect on household protein intake or
that there is lower availability of alternative protein sources, such as market-bought
fish. Alternatively, this result may be an artefact of lower enforcement effort in more
remote villages. Together, these results demonstrate the relative importance of spatial
effects on the number of meat meals consumed per household as opposed to those
associated with belonging to different socio-economic groups. Overall, a complex
pattern emerges, with the number of wild meat meals consumed determined by
availability (living within the forest), the time available to households to spend
hunting or fishing, the resources available to each household and opportunity to
choose alternative meat sources. Consequently, whilst the number of wild meals eaten
appears to have declined, it is difficult to attribute this decline to a reduction in stock
levels. However, as the total number of meat meals eaten from any source has also
decreased slightly, it is possible that there has been a decline in stocks that households
have not been able to replace.
Perceived household resource security
The results to the question on whether survey respondents felt secure regarding the
resources upon which they rely showed little variation between villages, with the
exception of O Chrar, O Rona and Sre Khtong (in which lower than average levels of
resource security were reported and which had all been recently affected by
concessions or illegal logging) and Pu Trom (in which higher than average security
was reported and which has an elephant sanctuary located close by with strong links
to the local community and ties to the conservation project). In total, 271 households
(44.9% of those that responded to this question) felt secure about their resources.
Correcting for population differences between villages gives a marginally different
adjusted figure for the whole project area of 47.6%. Of those households that felt
secure about their resources, the presence of conservation in the area was cited by
59.8% of respondents (Figure 3.6). For those that did not feel secure about their
resources, concessions (56.9%) and illegal logging (33.1%) were cited as the two
biggest threats.
A GLMM was constructed to understand which households felt most secure about
their resources, with the probability of feeling secure as the response variable (Table
50
3.8). All of the explanatory variables included in the final model were found to be
associated with feeling insecure (i.e. having a lower probability of feeling secure) and
were related to those households who rely on natural resources the most. Households
who collect NTFPs, have many resin trees or live in the core zone were found to feel
the least secure, all of whom are most reliant on natural resources. In many respects
this is unsurprising, as those with the most to lose could be expected to be most
concerned about the security of the resources upon which they rely, but it does show
that there is progress to be made with regard to increasing the security of natural
resources at the site.
Figure 3.6: Reasons given for why respondent households felt secure about their resources
Table 3.8: Coefficient estimates for the GLMM with the probability of feeling secure about
natural resources important for livelihoods. The difference in probability was found by a reverse
transformation of the logit link function used in the GLMM for an average household (i.e.
assuming sample average number of resin trees per household). Standard errors are given in
parentheses. Significance values: * = P < 0.05; ** = P < 0.01; *** = P < 0.001.
Variable
intercept
NTFP collector
no. resin trees
in core zone
Estimate
0.618 (0.054)
-0.210 (0.045)
-0.001 (0.000)
-0.145 (0.068)
51
P
**
***
*
*
Sufficient farmland to support livelihoods
Household land use
In 2007, the weighted average area of productive land claimed per household for the
villages sampled in both periods was 1.8 ha, with 70 households (20.2%) reported to
be landless. In 2012, the weighted average area of productive land for the same
villages had increased to 2.3 ha per household, with only 32 households (7.2%)
landless. Across the full 20 project villages, the weighted average land holding was
found to be 2.4 ha per household. Amongst those households with land, the average
reported area of land held increased from 2.2 ha to 2.6 ha per household between
2007 and 2012. However, it is likely that these figures do not reflect the true extent of
increases in land use (see Chapter 4).
A linear mixed model of the log-transformed area of land claimed per household was
constructed for land owning households (Table 3.9). The results of this model show
the variable with the greatest explanatory power in both time periods is the adjusted
household poverty score, with an increase of 0.1 in household poverty score found to
be associated with a 19.1% increase in household land. It is likely that this
relationship acts in both directions. Better off households are more able to acquire
additional land and have the resources to cultivate it, while acquiring more land is
likely to bring in greater earnings and, therefore, increase household economic wellbeing. A dummy survey year variable was not included in the final model, which
suggests that in real terms, despite increases in household economic well-being
between the two surveys, there has been no significant increase across the landscape
in the area of land held by households. As with the model of the number of resin trees
owned, both the marginal and conditional R2 statistics are low. Given that no other
correlation was found, this supports the argument that reporting of areas is unreliable.
Table 3.9: Coefficient estimates for LMM with the log transformed reported area of land claimed
per household as the response variable. Standard errors are given in parentheses. 𝑹𝟐𝒎 = 0.175,
𝑹𝟐𝒄 = 0.216.
Variable
intercept
adjusted poverty score
Estimate
-0.439 (0.095)
+1.908 (0.162)
52
t value
-4.608
11.767
The proportion of households that reported owning more than 5 ha was also virtually
unchanged between the two surveys, with 5.5% of households with more than 5 ha in
2007 and 4.8% in 2012. This suggests that the number of households complying with
community regulations or legal restrictions on land ownership may be relatively
stable, although it is difficult to give too much credence to these figures given that
interview respondents are unlikely to admit to rule breaking.
Land alienation
The results of the two household livelihood surveys show that the percentage of
landless households decreased between 2007 and 2012 (20.2% and 7.2% respectively).
Correcting for variation between villages by applying population weights gives
slightly revised figures of 19.8% landless households in 2007 and 8.3% in 2012.
These figures suggest that complete land alienation is less of a problem than before. A
generalised linear mixed model was constructed to investigate the probability of
households being landless in the two surveyed time periods in more detail (Table
3.10). Household poverty score, whether the household owned resin trees and a
dummy survey year variable were all included in the final selected model. The
relationship between landlessness and poverty score has already been shown, and the
model confirms that households with lower poverty scores have a higher probability
of being landless than those with higher scores. Resin tree owning households were
also considerably less likely to be landless. This is unsurprising, as owning resin trees
is closely associated with household well-being. For a household of average poverty
score, the probability of being landless fell by 0.15 between the two surveys,
suggesting that progress has been made towards reducing the extent of land alienation
in the study site. This is particularly heartening given that a programme of
participatory land use planning has been the one of the major conservation activities
implemented over this period.
Table 3.10: Coefficient estimates for the GLMM with the probability of a household being
landless as the response variable. The difference in probability was found by a reverse
transformation of the logit link function used in the GLMM for an average household (i.e.
assuming sample average household poverty score). Standard errors are given in parentheses.
Significance values: ns = P > 0.05; *** = P < 0.001.
Variable
intercept
adjusted poverty score
Estimate
0.214 (0.083)
-0.039 (0.001)
53
P
ns
***
resin tree owner
year (2012)
-0.143 (0.024)
-0.154 (0.019)
***
***
In 2012, interview respondents were also asked if they had lost any land to companies
in the last 5 years. Ten households (1.6%) from seven villages reported losing land, at
an average of 3.3 ha each. As with resin tree losses, the magnitude of these losses is
unsubstantiated but suggests that households remain at risk of losing land to
companies granted economic land concessions.
Discussion
In general, the overall status of the indicators considered in this chapter is largely
encouraging, particularly the improvements in household economic well-being
observed generally and for all vulnerable groups. Whilst some groups continue to be
at a disadvantage relative to other households, this has not prevented them from
experiencing an overall improvement in economic well-being, as has been reported
elsewhere in Cambodia (Clements et al., in press). The only caveat to this picture of
landscape-wide development is for households located in the most remote parts of the
project area, where distance to market centres offset the gains in economic well-being
observed elsewhere. It is expected, however, that the economic well-being of these
households will be found to have improved at the next iteration of the quantitative
livelihood survey, as access to this part of the project area is improving year by year.
Of the two social targets examined, progress has been made in relation to aspects of
both of them. For the target to increase natural resource security, the results relating to
the three of the main indicators considered suggest, on balance, that the key resources
upon which people’s livelihoods depend continue to be protected. Whilst the
probability of a household owning resin trees has fallen marginally, this can largely be
attributed to the increasing number of households living in the study site and the
number of resin trees owned by those households that have trees appears to be stable.
The number of wild meat meals per week has fallen but it is not clear whether this is
associated with increased availability of alternative protein sources or a decline in
stock levels. Perceptions of resource security were not recorded in 2007 so a
comparison between the two years was not possible. However, nearly half of all
households feel secure about their resources, of which the majority attribute this
feeling to conservation activities.
54
For the second target of ensuring sufficient farmland to support household livelihoods,
progress was more clear-cut. Most positive was the significant reduction in the
percentage of landless households, a group shown to have lower household economic
well-being than those with at least some productive land. As one of the main
interventions undertaken by the project in this period has been a programme of
participatory land use planning and indigenous land titling, in which the indigenous
communities of six villages have now been granted communal land title, this trend
demonstrates encouraging progress towards the project target (Evans et al., 2012b).
This is particularly encouraging given the high rates of landlessness across the
country (UNDP, 2011). The average land held per household was less encouraging, as
this does not appear to have changed. However, there are issues with using reported
areas (as discussed below) and it is thought likely that the area of household land has
in fact increased (see Chapter 4).
Of the indicators investigated, most concerning are project-wide population trends
and the percentage of households that feel insecure with regard to key natural
resources: land, resin trees and forest products. Whilst the majority of the rise in
household numbers has been centred on the lowland area close to the Keo Seima
district town of O Am, other parts of the project area are experiencing higher than
average growth. Immigration is a particular concern, as it is often cited as a reason
why households feel insecure regarding their natural resources and is a significant
driver of forest loss nationwide (Chan, 2008; McMahon, 2008; Biddulph, 2011).
Efforts to curb immigration (mostly centred on the enforcement of protected area
boundaries and community land use planning) are on-going but current trends remain
a concern. So too does the percentage of households worried about their resources. In
this case, the majority of households cited economic land concessions and illegal
loggers as the main threats. It is expected that project efforts to protect resources from
both of these threats will improve as a result of accreditation of the SPF REDD+
demonstration project and, hence, the percentage of households concerned for their
resources is expected to fall. However, the level of perceived insecurity in relation to
key natural resources is a concern, as it is throughout the country (Chan, 2008;
Grimsditch & Henderson, 2009; Biddulph, 2011).
55
Data issues
Whilst the picture presented by the results is generally positive, the quality of the data
used must be taken into account. Many of the indicators considered were selected in
such a way to minimise issues that can arise in areas where education and literacy
rates are low and responses highly variable. For instance, one aspect of household
livelihoods that is of particular interest is the income households receive from
different livelihood activities. Whist data for this was collected in both the 2007 and
2012 surveys, collection of income data can be subject to multiple issues, particularly
if long recall periods are used (Bernard et al., 1984). Consequently, household
poverty score was preferred for making comparisons in household economic wellbeing between the two time periods. However, it was not possible to completely
eliminate some of these sources of uncertainty, such as area estimation and short-term
recall.
Area reporting is particularly difficult. The majority of plots are irregularly shaped
and are often located on steeply sloping ground that makes estimation problematic,
even for experienced individuals and despite efforts to improve the quality of area
data collection. During the implementation of the 2012 quantitative livelihood survey,
trained CENTDOR enumerators paid particular attention towards assisting
respondents in answering area questions as accurately as possible. In some cases, area
estimation may also have been improved by the manner in which households sell their
cassava. Often cassava yields are now sold on the basis of field area and an
assessment of quality, not on actual yield, with cassava traders supplying the labour to
harvest whatever the field yields (S. Milne, pers. comm.). Fields are measured as part
of this process, which may aid some farmers in becoming more aware of their field
sizes. These cases are the exception, however, and so area estimation remains an issue.
In addition to the difficulties in estimation, there are also concerns that farmers may
choose to deliberately misreport areas, particularly if they have cleared forest illegally
outside of agreed community boundaries or have exceeded area limits. Analysis
presented in Chapter 4, for which household plots were measured with GPS units,
found significantly higher household land areas than those reported during the 2012
livelihoods survey in the two villages sampled as part of this analysis. This is a
concern, and it shows that more rigorous methods are required to collect data relating
to the area under cultivation for each household.
56
Possibly the largest source of error in the data presented does not come from
misreported values but from activities that were not covered by the 2012 survey for
reasons of cultural sensitivity and because the CCB manual for REDD+ impact
assessments states that illegal activities do not need to be considered when assessing
the social impacts of projects (Richards & Panfil, 2011). In landscapes such as SPF,
where involvement in small-scale illegal activity is high, this can lead to problems in
interpreting the data. For example, if the project is successful in reducing illegal
logging or forest clearance (both important targets with respect to REDD+), it is
conceivable that household economic well-being indicators will reveal a net decline.
Certain vulnerable groups, who may become involved in illegal activities for lack of
alternative options, are likely to be disproportionately affected. In this situation, there
are potentially complex trade-offs and synergies between conservation objectives.
Effective action towards one project target (reduced deforestation or degradation) may
seem to penalise progress towards another (increased economic well-being) for some
groups (e.g. households involved in illegal logging) but benefit others (e.g. resin
tappers). These trade-offs may even occur within the same household (i.e. a
household may benefit from improved protection of resin trees but lose out if they are
involved in illegal logging or land clearance). Under CCBA guidance, however,
negative impacts from illegal activities can be ignored, but it is difficult to attribute
reductions in household economic well-being to successful project activities without
monitoring involvement in illegal activities. Complementary qualitative monitoring
approaches, which can significantly improve understanding of research contexts
(Drury et al., 2011), may serve to minimise the confounding effect of economic gains
from illegal activities but it cannot be assumed to be eliminated entirely.
Indicator selection
One of the principal objectives of the 2012 quantitative livelihood survey was to test
the appropriateness of the indicators selected as part of the social monitoring
framework. The monitoring of social impacts resulting from project activities in a
context such as SPF, where livelihood strategies are going through a period of
significant change as a result of improved access and greater integration with the
market economy, is highly complex, which makes it difficult to select indicators that
reflect changes that are directly attributable to project activities or threats (as
evidenced, for example, by the difficulties in interpreting trends in wild meat
57
consumption). However, identifying suitable alternative indicators is also difficult.
For example, in order to assess progress towards the target of ensuring sufficient
farmland to support household livelihoods, an understanding of how much land is
needed to be sufficient is first required. This is problematic, since both agricultural
productivity and livelihood strategies vary. A household that grows cash crops may
require less land than one that grows rice. In this instance, it may be sufficient to ask
each household if they are able to feed themselves without difficulty all year round,
but subjective questions such as this are vulnerable to bias or strategic responses.
Consequently, one of the key pieces of information required to assess progress
towards this project goal remains illusive.
This difficulty in selecting alternative indicators or indicators which provide clear
evidence of positive or negative trends undermines the suitability of applying the
theory of change based approach to impact assessment recommended by the CCBA.
Although this approach has advantages in comparison to other more robust
approaches, such as the use of matched samples (e.g. Arriagada et al., 2012; Clements
& Milner-Gulland, in press) or randomised trials, it can be difficult to accurately
assess impacts in contexts that are subject to strong external drivers of change in the
absence of a counterfactual. Employing complementary qualitative methods, which
would allow the project team to discuss the causes of observed trends with local
communities, may reduce the challenge of interpretation. In addition, further
refinement of some of the indicators included in the monitoring framework may be
required. It is recommended that the causal chains presented within the conceptual
model (Figure 3.1) are refined following the completion of the proposed qualitative
discussions. In this way, the interpretation of changes in the status of different
individual indicators can be improved and placed within the framework of each causal
chain. For example, the interpretation of indicators related to the target of improved
resource security should be considered in the context of changes to indicators for both
the relevant direct and indirect threats identified in the causal chain for this target. If
indicators change in the expected direction at each stage along the causal chain,
greater confidence can be given to the resulting conclusions.
58
Chapter 4
4. A tale of two villages: an investigation of conservationdriven land tenure reform in a Cambodian Protection Forest
Introduction
Sparsely populated, resource rich forest habitats have frequently been subject to
exclusionist policies by former colonial or national governments, repeatedly failing to
recognise the rights of the people living inside such areas (Colchester, 2004).
Biodiversity conservation has a particularly chequered past in this regard, with the
protectionist ‘fences and fines’ approach dominating conservation practice throughout
much of its history (Adams, 2004). Given the high biodiversity value of many forest
habitats and often weak political representation of forest peoples, it is no surprise that
this approach has brought conservation practitioners and forest inhabitants into
conflict (Brockington & Igoe, 2006). More recently, Government efforts to meet
commitments made under the Convention of Biological Diversity (CBD) to set aside
areas for the protection of biodiversity have brought further accusations of ‘green
grabbing’ (Fairhead et al., 2012).
From the 1980s onwards, however, there has been increased recognition of the
customary rights of indigenous peoples (IPs), with both international and national
legislation slowly moving to reflect this. This has been reflected within conservation,
with article 8(j) of the CBD, decision 7.23 of the 2005 CBD Conference of the Parties
and the CBD's 2010 Aichi target 18 all requiring signatories to respect the rights and
practices of indigenous and local communities. Beyond international agreements,
there has been increasing acceptance within conservation organisations that the rights
of local communities must be considered, which has resulted in a movement towards
the principle of ‘doing no harm’ as a minimum requirement (Adams et al., 2004).
This change of approach is well reflected in the increasing trend towards the adoption
and recognition of various forms of indigenous or community conserved areas
(ICCAs). As of 2008, it was estimated that ICCAs encompassed over 400 million ha
in 28 of the world’s 30 most forested states, a significant increase on similar
59
calculations from 2002 (Sunderlin et al., 2008). This represents a significant step
forward in the recognition of customary tenure rights, although rights-related issues
are often not the primary impetus behind the adoption of ICCAs (Berkes, 2009). One
argument in support of ICCAs lies in their perceived effectiveness in providing
greater protection benefits in comparison with more traditional forms of conservation
management. Whilst there is no guarantee that ICCAs result in positive biodiversity
outcomes, there is some evidence that points in this direction (for a review see
Shahabuddin & Rao, 2010). Furthermore, as payments for environmental services
schemes become increasingly widespread, clarification of disputed or informal tenure
arrangements becomes ever more necessary to ensure that payments are effective and
customary rights holders do not lose out (Engel et al., 2008; Milne & Niesten, 2009;
Larson, 2011).
In part, the success of rights-based approaches is dependent on the manner in which
they are implemented. The rise of ICCAs was preceded by greater interest in
participatory approaches to conservation (Agrawal & Gibson, 1999; Berkes, 2004)
and in this regard participatory land use planning approaches have gained
considerable support. Such approaches are thought to be an improvement on topdown land use planning methods, ensuring that the process is more inclusive, makes
greater use of local knowledge, reduces the risk of future disputes and supports local
level institutional development (McCall & Minang, 2005; Cronkleton et al., 2010).
Concerns have been raised, however, regarding the effect of participatory land use
planning approaches on local land use and the degree to which land access ultimately
reflects power arrangements within a village. In an investigation of pilot participatory
land use planning projects in Laos, Lestrelin et al. (2011) found evidence to suggest
that, despite village participation, the process resulted in maintenance of the status
quo, with the extent of individual households’ access to land dependent on their
power to negotiate with village elites. Despite participatory land use planning
approaches being widely applied in multiple contexts and for different purposes,
published case studies are rare and outcomes are often left unevaluated (Bourgoin,
2012). Where case studies have been published, there is a distinct lack of analyses
looking at how different groups within villages have been affected by and comply
with the institutional changes brought about by participatory tenure reform
instruments.
60
In this chapter, I seek to redress this gap. I evaluate the implementation of indigenous
land titling, a product of a participatory land use planning process, in two villages in
the Cambodian uplands, investigating at the household level how land use has
evolved following the initiation of this process. I do this in the context of a dynamic
but varying socio-economic landscape that is driving significant land use and
demographic change. In this way, it is possible to compare the performance of tenure
reform in two villages that have been exposed to different levels of social and
economic change.
An analysis of household land use change and compliance with both the legally
binding zonation of community lands and community-agreed rules governing land use
within community titled areas is presented. The land holdings of individual
households had been measured at the beginning of the titling process in each village.
These were re-measured for a sample of households in each village and compared
with the size, location and use of the holdings to that recorded as part of the process.
These data were used to evaluate the extent to which the indigenous customary lands
and the agreed conservation areas of the two communities have been successfully
protected from land conversion from forest to agriculture. The extent of individual
household land use change since the start of the land tenure reform process was
analysed, in the context of rapidly changing social and economic circumstances (in
particular, immigration of poor landless households). Within this analysis, I
investigate the effects of different socio-economic characteristics thought to affect
land use decision-making, including ethnicity, available household labour, livelihood
options, wealth and residency, on the behavioural response of households to land
tenure reform. Household understanding of the tenure reform process and perception
of tenure security in each of the study villages were also examined. This provides a
valuable case study from which to draw lessons, not only for the implementation of
land tenure reform for Cambodia’s minority IPs, but for participatory approaches to
land use planning more widely.
Indigenous land tenure reform in Cambodia
Cambodia is home to several minority indigenous peoples. Although customs and
practices vary between groups, Cambodia’s minority IPs have typically operated
under a traditional system of collective customary land tenure with a strong spiritual
61
connection to the surrounding landscape, particularly forests (Baird, 2000; ADB,
2002; Baird & Dearden, 2003; Fox et al., 2008; Simbolon, 2009). Traditional
livelihood systems were largely based on small-scale swidden agriculture, with only a
small proportion of communal lands cultivated at any single point in time. In recent
years, however, there has been a shift in many communities away from traditional
agricultural practices, with increasing reliance on the production of commercial and
tree crops (Fox et al., 2008).
Largely unaffected by the private property regime of French colonial rule,
Cambodia’s IPs came under increasing pressure to integrate into Khmer society after
independence, culminating in forced integration and mass resettlements during the
Khmer Rouge period (1975-1979). Following Vietnamese invasion in 1979,
Cambodia’s minority groups were allowed to return to their ancestral lands, although
many did not return until the 1990s. After the 1991 peace accord, a series of laws
aimed at reforming land ownership was enacted, most notably the 2001 Land Law.
From the perspective of indigenous communities, the 2001 Land law is particularly
important as it provides the legal basis to secure customary land rights as well as
safeguarding these rights until such a time as the legal title is granted. Once title is
granted, these lands are classified as state private lands, meaning that they remain the
property of the state but have no public interest (Oberndorf, 2005). As such,
communities do not hold the right to dispose of their land. In order to apply for
indigenous communal tenure (ICT), a community must first register as an indigenous
community (IC) and be recognised as such by the Ministry of the Interior. Progress
towards ICT has been slow, with only eight villages having received title at the time
of writing (a further 190 are planned; Milne, 2013).
Despite legal reform recognising the rights of IPs and rural smallholder framers, land
disputes, alienation and large-scale land grabbing are frequent and widespread. While
large-scale land acquisitions are by no means a recent phenomenon in Cambodia, the
situation has been greatly exacerbated in recent years. The human rights NGO
ADHOC reports that 2,657,470 ha (approximately 17% of the total land area of the
country) had been granted as economic land concessions (ELCs; areas of up to 10,000
ha granted to industrial companies for intensive agriculture), as of December 2012
(ADHOC, 2013). Not only has this had extreme implications for Cambodian
62
smallholder farmers but also for conservation. In 2012 alone, ELCs covering 381,121
ha were granted, of which over 70% are within existing protected areas. Given the
high degree of overlap between Cambodia’s indigenous peoples and the country’s
protected area network, efforts to secure the customary rights of IPs may also serve as
added protection from ELCs for protected areas.
Land tenure reform in the two villages
The research presented in this chapter was undertaken in two villages, Andong
Kraloeng and O Rona, in Mondulkiri Province, northeastern Cambodia (Figure 4.1).
Both villages are located within Seima Protection Forest (SPF; Chapter 2). The
predominant minority indigenous group in the area are the Bunong, who belong the
Mon-Khmer language family (Bourdier, 2009). During the Khmer Rouge period, the
area was almost totally abandoned, as households living there were forcibly resettled
in the lowland north of Mondulkiri Province, with the majority of families returning
in the 1990s and early 2000s. The resulting loss of knowledge regarding customary
land use led PA authorities to initiate a programme of working with indigenous
communities to map their customary use areas and to plan future land use. This
programme, which was piloted in Andong Kraloeng in 2003 and later expanded to
include other villages, involved a series of consultations with local communities,
following procedures based on the government’s manual for participatory land use
planning (Rock, 2001).
Andong Kraloeng is a Bunong village, consisting of six smaller sub-village
settlements, located within the densely forested core zone of SPF on the main road
between Phnom Penh and the provincial capital, Sen Monorom. Despite being
situated along the main road, it has been little affected by immigration from other
provinces. In 2011, a sub-decree was passed recognising the rights of the Andong
Kraloeng IC, making them the third IC nationally to be granted rights over their
customary land (UNOHCHR, 2011). The second study village, O Rona, is located at
the edge of SPF and the adjacent Snoul Wildlife Sanctuary, managed by the Ministry
of Environment (MoE), and consists of five smaller sub-village settlements.
Historically, the village was an indigenous Bunong community but in recent years it
has been heavily affected by an influx of immigrant Khmer families seeking land. As
the village is situated close to both the district capital and the border with Vietnam, it
63
is considerably more integrated into the market economy than Andong Kraloeng. This
has affected traditional livelihoods, with a greater reliance on commercial agriculture
readily apparent amongst Bunong households. It also presents a greater threat to
community lands from speculators and immigrants; at the time that the land tenure
reform process was initiated in 2005, there were 35 claims to land inside the
traditional village boundary from households living in other villages.
Figure 4.1: Map of Seima Protection Forest (2,927 km2), showing the location of the two study
villages.
At the start of the ICT process in both villages, a provisional Indigenous Community
was set up. Under Cambodian law, formal recognition by the Ministry of the Interior
of a community’s status as an indigenous community is a legal pre-requisite for
applications for ICT. For each IC, PA authorities facilitated the selection of a
management committee, with committee members drawn mainly from traditional
village elders. Every indigenous household living in the village at that time elected to
join. Following this, the former extent of customary use was investigated using a
variety of sources (including historical topographical maps, aerial surveys, forest
inventories and local knowledge) and all land holdings within the provisional village
area were mapped by PA authorities. On completion, a series of areas was created
delineating different land uses (Table 4.1). These included areas designated for
current and future agriculture, as well as community forest in which the only form of
64
resource use permitted is the collection of non-timber forest products (NTFPs). These
zones are managed by the IC committee.
Table 4.1: Details of the participatory land use planning process in the two study villages. Land
inside areas designated for agriculture is a mix of cultivated land and forest that, as yet, has not
been cleared.
Andong Kraloeng
land use designations:
residential/agriculture
NTFP forest
spirit forest
burial forest
total village area
O Rona
1,323 ha
23,584 ha
43 ha
27 ha
25,003 ha
1,477 ha
2,274 ha
44 ha
3,795 ha
2003
2008
2010
2011
2005
2009
Dec 2012
2013
Population (heads):
start of PLUP process
June/July 2012
390 (2004)
563
573 (2006)
1041
number of households:
start of PLUP process
June/July 2012
89 (2003)
124
121 (2006)
229
IC formed/land mapped
ICT application
ICT zones redrawn
ICT granted
In addition to the creation of the different land use areas, the ICT process supported
the agreement of a series of rules governing household land use, made enforceable by
traditional village sanctions, to assist in the management of village land. These
regulations, which were largely drawn up from informal traditional practices, were
designed to allow the IC to plan for future growth but also to protect traditional
livelihoods. As such, limits were put on the area of paddy land and tree crops allowed
for each household. A village constitution, which details the formal composition of
the committee, the goals of the IC and criteria for IC membership, was also drawn up.
The main difference between the procedure followed in O Rona and that of Andong
Kraloeng was the number of Khmer migrant families living in the village at the time
of the ICT process. Under the 2001 Land Law, households are entitled to claim
ownership of occupied land provided their claim had been uncontested for at least five
years prior to the promulgation of the law. Land claimed after 2001 is not eligible for
private ownership and under law remains state property. Although the ICT process in
O Rona included every household living in the village, Khmer families were only
65
permitted to retain the land they already occupied. Any expansion of their land within
those areas identified for future agricultural use would be illegal under the 2001 Land
Law. This created a two-tiered system within the village, with Khmer immigrant
households granted fewer rights than their returnee Bunong neighbours. This was not
an issue in Andong Kraloeng as there were no Khmer households in the village at the
time of the ICT process. In both villages, the rules governing land use within the
village area apply only to Bunong households and to the Khmer households present
during the ICT process, as recent settlers are not permitted to claim any land within
community titled areas. Newly formed households of married children of IC members
automatically become members of the IC, with the accompanying land use rights and
responsibility to abide by community regulations.
Methods
Fieldwork for the research was carried out in both villages between May and July
2012, and consisted of a series of focus group discussions, household structured
interviews and land use mapping.
Focus group discussions
The aim of the focus group discussions was to gain a better understanding of land use
practices and to investigate local perceptions regarding land issues. Each focus group
consisted of 8-15 participants and lasted approximately two hours. In each village,
one focus group was held with members of the IC committee responsible for the
management of community area. In O Rona, three additional focus group discussions
were held, which separated participants who had been present at the time of the initial
planning exercise and those who had not, as there is a strong distinction in rights
between newly arrived and resident households. In Andong Kraloeng, immigration
rates are low and the population relatively homogeneous, so this was not deemed
necessary.
Household interviews and land use mapping
Structured household interviews and land use mapping were conducted for 114
households, with 44 sampled in Andong Kraloeng and 70 in O Rona (approximately
one third of the households in each of the study villages), stratifying proportionally by
sub-village settlement and randomly sampling within each stratum. No households
66
living outside either of the villages were sampled. This approach provided a crosssectional sample and was selected to ensure that both newly arrived immigrants and
newly formed households (e.g. through marriage) were included in the study. As the
original land use mapping process included every plot of land within each community
zone, a cross-sectional survey design provided sufficient matches between the two
time periods to allow for simple panel analyses.
The purpose of the structured interviews was to collect socio-economic and land use
data for each household, as well as gathering information on perceptions of land
issues and knowledge of the ICT process. Where possible, all interviews were
conducted with the head of each household and lasted for approximately 45 minutes.
Following the completion of each interview, all land parcels held by the respondent
household were measured by walking the parcel edge and recording the path taken on
GPS. Observations were made regarding the crops grown, the likely age of those
crops and adjoining land uses and owners to triangulate information given during
interviews.
Land use mapping check
A set of 100 random points was sampled across each village area, and the owner of
the land at each point identified, in order to check whether information about sensitive
plots was being withheld during the interviews. Given that a significant proportion of
households hold land illegally, it was considered likely that participants might opt to
withhold sensitive plots. As sampling within each village was incomplete and did not
cover land claimed by outside interests, uncertainties existed regarding whether the
team had been shown all plots used by the households sampled. Recent land cover
analysis (WCS, 2013) and satellite imagery were used to identify areas under
cultivation. Random points were generated within un-visited areas of cultivated land
for each settlement using ArcMap version 10.0. Local guides or members of the
village ICT committees assisted in identifying the landowner at each point. The
names collated were then cross-checked against the list of households already
interviewed to ascertain whether any plots had been withheld. This provided a
measure of the number of withheld plots for each village. In cases where plots had
been withheld, the household head was re-interviewed and the plot mapped.
Interviewers were careful at this stage to stress that they were clarifying the details
67
given during the original interview and to avoid any inference that the respondent had
deliberately misled the research team.
Spatial compliance analysis
Land use compliance was measured against two sets of criteria: with the law and with
IC regulations governing land use within community boundaries. Plots measured in
2012 as part of this research were compared with those of each household as
measured during the previous ICT process and with the various boundaries created
through the ICT process. In certain cases, records were incomplete, which had the
effect of reducing the overall sample size for the compliance analyses to 105
households. These spatial comparisons were carried out using Quantum GIS version
1.8.0. A margin of error of 0.05 ha or 5%, depending on which was greater, was used
to allow for differences in digital and physical boundaries.
For Andong Kraloeng, these analyses were made more complicated as boundaries
were changed in 2010. This created a situation where plots that had previously been
compliant were made non-compliant and vice versa. In the former case, plots were
considered to be compliant for the purposes of this analysis, despite the boundary
changes. In the latter case, plots were considered non-compliant if they were outside
community agricultural areas when initially cleared, but compliant if they were within
community agricultural areas when cleared. For both cases, it was necessary to use
reported plot age, triangulated against observations made during plot measurement, to
check whether the plot had been cleared at the time the boundaries were changed. For
O Rona, the boundaries had not been changed at the time of the survey so this was not
an issue. In both villages, however, it was necessary to correct for whether or not each
household was a member of the IC. In Andong Kraloeng, all non-members of the IC
were excluded from holding land within the community area, but at the beginning of
the ICT process all households were members of the IC. In O Rona, non-IC
households that were present in the village at the beginning of the ICT process were
allowed to keep the land they owned at that time but any further expansion was
prohibited, while households that had moved to the village afterwards were excluded
from claiming any land.
In addition to analysing land use compliance with the law, compliance with five
regulations governing land use within IC managed areas was also checked. These
68
were regulations banning: 1) buying, selling or renting land, 2) clearance of spirit or
burial forest, 3) exceeding 2 ha of land for tree crops (such as cashew or rubber) per
household, 4) exceeding 1 ha of land for cultivation of paddy rice per household and
5) exceeding 5 ha of land under any form of cultivation per household. In each case, if
these regulations had been broken prior to the rules being drawn up, then the
household was considered compliant provided that there were no new infractions. For
example, if a household cultivated 7 ha of land prior to the start of the ICT process,
they were allowed to keep all 7 ha. If, however, they subsequently cleared more land
within the community area, they were judged to be non-compliant with the 5 ha area
limit. For the purposes of this analysis, only land within the community titled areas
was considered, as the IC regulations only govern land use within these areas.
Statistical modelling
Linear and generalised linear mixed models (LMMs; GLMMs) were selected using
backwards step-wise selection methods based on the small sample size corrected
Akaike Information Criterion (AICc; Akaike, 1974; Burnham & Anderson, 2002)1.
Models with the lowest AICc were selected except in cases where a more
parsimonious model was found with a ∆AICc value of less than two. The variable
settlement, which referred to each sub-village settlement surveyed, was included as
the only random effect in all models (see Appendix B; Table B.2 for a list of the
explanatory variables considered for each model).
In order to model the area of household land claimed illegally, a hurdle modelling
approach was used to account for the high number of zero values (Mullahy, 1986).
Compliance with the 2001 Land Law was modelled first using a simple binary
variable. Subsequent modelling of correlates of illegal land area claimed was carried
out only for those households with some illegal land. All area variables were
transformed using natural logarithms. In cases where area variables contained zero
values, a constant equal to half the lowest non-zero value was first added to all data
points. The exception to this was for the model of total area claimed per household.
For this model, three zero value data points were removed and the sensitivity of the
1
All statistical analysis was carried out using R version 2.15.1, on RStudio version 0.97.314. Within R,
the lme4 package version 0.999375-36 was used to analyse all models.
69
coefficients tested. Removing these data points did not have a significant effect on the
model estimates and greatly aided interpretation of the results.
In each case, once the final model had been selected, visual validation was conducted
to check for residual normality, heteroskedasticity and possible correlations between
fixed effects and the residuals. Over-dispersion in binomial logistic regression models
was checked by comparing the sum of squared Pearson residuals with the
approximate residual degrees of freedom. No issues were found in any checks.
Marginal and conditional R2 statistics for LMMs were calculated following Nakagawa
& Schielzeth (2013).
Results
Land Use Change in the Two Villages
Andong Kraloeng
In Andong Kraloeng, 44 households were interviewed, of which 43 currently claim
land (for plan of mapped fields see Appendix B; Figure B.1). The area claimed by
these households totalled 133 ha, with 119 ha in cultivation. This represents a
significant expansion in the average area of land under cultivation per household,
from 1.0 ha in 2004 to 2.7 ha in 2012 (Table 4.2). Extrapolating this land use for the
entire village population gives a total area under cultivation of 334 ha, which remains
a small fraction of the 1398 ha granted to the IC for agricultural and residential
purposes, and shows that there remains considerable scope for agricultural expansion.
A shift away from traditional indigenous land use systems was observed. Such
systems tend to be highly diverse, with many different crops commonly grown in
small amounts around the edge of the same parcel of land (Baird & Dearden, 2003),
but the discussion here is limited to changes in the dominant crop in each field. The
most dramatic change observed is the transition to commercial agriculture. In 2004,
the average area under some form of cash crop cultivation (principally cashew) was
0.6 ha per household, or 60% of the land under cultivation. Much of this land,
however, was also used to grow non-cash crop upland rice intercropped with the main
cash crop. For instance, 0.5 ha per household, equivalent to half of all land use within
the village, was used for cashew plantations with rice grown underneath or between
70
young trees. On average, 85% of cultivated land had rice as the dominant or
intercropped crop. By 2012, the average area of cash crops had risen to 2.5 ha per
household, or 93% of the area cultivated, making the increase in cash crops greater
than the observed increase in overall land use. The majority of this increase is
accounted for by the emergence of cassava, a crop that was unrecorded in 2004. In
2012, the average cultivated area of cassava was 1.8 ha per household (67% of the
cultivated area), with 1.4 ha per household intercropped with cashew. In contrast, the
importance of rice had dropped, with only 0.2 ha (7% of the cultivated area) per
household cultivated with rice as the dominant or intercropped crop.
Table 4.2: Mean area [ha] given to different land uses per household for the two villages at the
time fields were mapped by PA authorities as part of the ICT process and in 2012. The
percentage of average household area given to each land use is shown in parentheses. As
intercropping is common practice, percentages do not sum to 100.
Land use
all cultivation
cash crop (cassava,
cashew and rubber)
tree crop
rice
cassava
Andong Kraloeng
2004
2012
1.0
2.7
O Rona
2006
3.1
2004
4.8
0.6 (62)
2.5 (93)
2.9 (94)
4.2 (88)
0.7 (68)
0.9 (85)
0.0 (0)
1.4 (50)
0.2 (7)
1.8 (67)
2.8 (90)
0.8 (26)
0.1 (3)
1.9 (40)
0.6 (13)
3.0 (63)
Not only does this show an increasing reliance on producing cash crops, but it also
shows a breakdown of traditional rotational farming. As of 2012, cashew (a tree crop)
was present in approximately 50% of cultivated land, much of which was formerly
used for growing rice. As cashew is considered to have a 20 year productive life span,
land that would formerly be returned to fallow, and therefore collective ownership, is
now being retained as individually held land (albeit within communal title) through
the planting of cashew.
O Rona
The average land holdings in 2012 for the 70 households sampled in O Rona was 4.8
ha (for plan of mapped fields see Appendix B; Figure B.2), significantly more than in
Andong Kraloeng. This represents an increase of 1.7 ha (or 55%) per household from
2006, when fields were mapped as part of the ICT process. As with Andong Kraloeng,
these figures demonstrate a rapid expansion in the land being cultivated. There was a
significant difference observed in the area of land claimed between households that
71
were part of the IC and those that were not. For households that were sampled in both
2006 and 2012, members of the IC claimed on average 6.3 ha in 2012 (up from 4.2 ha
in 2006), whereas non-IC members claimed only 3.5 ha in 2012 (up from 2.5 ha in
2006). This, however, represents a similar rate of growth for both types of household.
Commercial agriculture was already widely practiced in 2006 in O Rona, with some
form of cash crops grown on 94% of agricultural land. In 2012, the proportion of
agricultural land involved in commercial cropping was essentially unchanged (88% of
cultivated lands), although the absolute area was much increased. The main observed
difference between the two time periods was the change in intercropping practices
between commercial and subsistence crops. In 2006, 59% of agricultural land (1.8 ha
per household) had a commercial crop as the dominant crop but was intercropped
with rice. This had fallen to just 5% of cultivated land (0.2 ha per household) by 2012.
Overall, the area per household on which rice was grown had contracted slightly from
0.8 to 0.6 ha per household. In contrast, though, the area on which it was grown as the
dominant or sole crop had increased from 0.2 to 0.6 ha per household, largely through
new areas suitable for paddy farming being cleared. This shows an increasing
delineation between areas assigned to growing cash crops and to subsistence rice
cultivation.
As with Andong Kraloeng, the most dramatic shift in O Rona has been the uptake of
cassava, which was first introduced in 2005. Less than 0.1 ha per household was
being grown in 2006 but this had increased to over 3 ha per household by 2012. This,
in part, is driving the changes observed in rice growing practices, as cassava is often
sold to traders by the field. Households, often those short of labour, agree a sale price
set by the area of their field and quality of the cassava, as opposed to harvesting and
processing the cassava themselves. Consequently, it makes sense for households to
have separate plots for different crops, even within the same field.
The emergence of cassava has also changed the area given to tree crops. In 2006, the
average household had 2.8 ha of tree crops, largely intercropped with rice, but by
2012 this had dropped to 1.9 ha and was intercropped with cassava instead. This
demonstrates a trend reported during household interviews that some farmers have
chosen to replace areas previously planted with tree crops with cassava. Overall,
however, the reduction in area planted with tree crops does not indicate a return to
72
traditional rotational cultivation as cassava is being grown year after year, despite
fears of declining yields.
Household land holdings
The selected model for the total log transformed area of land claimed by each
household in 2012 included three significant explanatory variables (Table 4.3): the
age of the head of the household, whether or not they had arrived after the ICT
process was initiated and an adjusted poverty score based on the basic necessity
survey methodology (Davies, 1997; see Appendix B for an explanation of how this
score was calculated). This meant that neither the household's livelihood options nor
available labour had a significant effect on land holdings.
Table 4.3: Parameter estimates for the household land holdings LMM, with ln(area) as the
response variable. One random effect was included in the model: settlement. The full list of
variables modelled, plus their explanations, are given in Appendix B; Table B.2. Standard errors
are given in parentheses. 𝑹𝟐𝒎 = 0.382, 𝑹𝟐𝒄 = 0.410.
Variable
intercept
age
poverty score
immigrant
Estimate
-2.365 (0.528)
+0.016 (0.006)
+4.166 (0.667)
-1.025 (0.194)
t value
-4.478
2.459
6.828
-3.132
There was a positive effect of age on household land holdings, predicting an increase
of 1.6% in the area claimed for every year increase in the age of the household head.
A much stronger correlation was found for the adjusted poverty score. In this case, a
0.1 increase in adjusted poverty score was found to result in an increase of 52% in the
total area claimed. This results in a predicted 272% difference in the total area of land
claimed between the best off and poorest families. Immigrant households, however,
were found to have smaller land holdings, with the model estimates predicting that
households that arrived after the ICT process had started had 103% less land on
average.
Household compliance
Compliance with the 2001 Land Law
Overall, there were high levels of non-compliance in both study villages, although
distinct differences in behaviour between the two were observed. In Andong Kraloeng,
73
26% of cultivated land had been cleared outside agreed boundaries, whereas the
figure for O Rona was 47%. Similarly, while 52% of sampled households in Andong
Kraloeng were found to have some illegal land, this figure was 79% in O Rona.
Comparing the performance of the indigenous communities within each village
reduces this difference slightly, with 49% of IC member households non-compliant in
Andong Kraloeng and 67% in O Rona.
In order to investigate compliance further, a GLMM, with a binary household
compliance variable as the response, was constructed to look at whether or not a
household claimed any illegal land (for a model summary table see Appendix B;
Table B.3). Only two explanatory variables considered for selection were included in
the final model; whether the household was part of the village IC and the natural log
transformed area claimed by each household. The difference in compliance between
the two villages was accounted for by the inclusion of the settlement lived in by each
household as a random effect in this model.
As the interpretation of logistic regression coefficients is not intuitive for log
transformed variables with constants added to account for zero values, the effect of
the area claimed by each household on the probability of compliance with the 2001
Land Law is plotted for both IC and non-IC member households (Figure 4.2). Among
non-IC member households, only those with very little land are likely to comply with
the law. This result is, perhaps, unsurprising, as non-IC members have little right to
land within either village. As the majority of sampled non-IC member households live
in O Rona, this result strongly reflects the two-tiered system that was created there. In
fact, the three non-member households that have not claimed land illegally hold no
land at all and provide labour for other households. None of the 20 sampled non-IC
member households that were present in the village in 2006 was found to be
compliant with the law in 2012.
Comparing the curves for IC and non-IC households reveals a strong positive effect of
being an IC member on compliance. This effect diminishes rapidly, however, as the
area of household land holdings increases. The predicted probability of being
compliant with the law for an IC member claiming the average area of land is just
over 0.2, illustrating the very low levels of compliance for families with average or
greater land holdings (Figure 4.2).
74
Figure 4.2: Predicted probabilities of compliance with the 2001 Land Law for non-IC and IC
households. The vertical dashed line shows the average area claimed per household for the whole
sample.
The second part of the compliance hurdle analysis comprised a linear model with the
log-transformed area of illegal land held by non-compliant households as the response.
Three explanatory variables were included in the final model (Table 4.4): the total
land holdings of each household, whether households were members of the IC and
whether the household head was Bunong. This final variable was possible because
several Khmer men had married Bunong women and moved into their wives’ home
villages. Under the rules drawn up by each IC, these households automatically
became IC members even though the head of the household was Khmer. The results
of this model again show a strong correlation between compliance and the total area
of land claimed by each household. In this case, an increase in the total area of 1%
resulted in a 1.1% increase in the area of illegal land held per household, suggesting
that households hold legal and illegal land in roughly equal proportions.
Table 4.4: Parameter estimates for the illegal household land holdings, with ln(illegal area) as the
response variable. P values significant at the 95% confidence level are shown in bold. The full list
of variables modelled, plus their explanations, are given in Appendix B; Table B.2. Standard
errors are given in parentheses. R2 = 0.613. Significance values: * = P < 0.05; *** = P < 0.001.
Variable
intercept
ln(area)
IC member
indigenous
Estimate
-0.359 (0.156)
1.084 (0.107)
-0.500 (0.246)
-0.521 (0.245)
75
P
*
***
*
*
More informative, from the perspective of evaluating the outcomes of the ICT process,
are the effect estimates for being an IC member at the time the ICT process was
initiated and for the household head identifying themselves as indigenous. The effects
of these two variables are comparable, with an approximate 65% reduction in the
predicted area of illegal land held in both cases. This suggests that, even among those
households that had held land illegally, there is a positive effect on compliance from
being a member of an IC. The second of these two results suggests that, holding the
effect of being an IC member constant, households with indigenous heads hold less
illegal land than those with heads from non-indigenous backgrounds.
Compliance with Community Regulations
The second aspect of household compliance considered was whether households
complied with the set of five regulations that were drawn up as part of the original
agreements in each village and which govern land use within community titled areas.
Overall, IC regulations experienced a higher rate of compliance than observed for the
2001 Land Law. In Andong Kraloeng, 77% of households were found to be compliant
with all five IC regulations considered. In O Rona, the figure was lower, with 52% of
households compliant.
In Andong Kraloeng, three of the rules were obeyed by every household sampled:
those prohibiting the buying and selling of land, clearing land in spirit and burial
forest and exceeding 1 ha of paddy land (Table 4.5). The regulation with the lowest
level of compliance was the limit on tree crop area, with all households who were not
totally compliant breaking this rule. This reflects the changes in the traditional
rotational cultivation system observed earlier. In O Rona, only the ban on clearance of
burial or spirit forest was completely observed (Table 4.5). In contrast to Andong
Kraloeng, nearly 50% of households were found to be breaking total area restrictions
within the community titled area. In part, this was due to non-IC member households
claiming land that they had not possessed in 2006. At 68%, the compliance rate for IC
members was still lower than in Andong Kraloeng, but higher than for the village
overall.
A mixed model approach to analysing IC regulation compliance was deemed
unnecessary in this case, as no difference in compliance was attributed to different
76
settlements. Instead, a generalised linear model (GLM) was produced, with three
explanatory variables included in the final selected model: the log-transformed total
household land holdings, whether the household was an IC member and whether the
household held some illegal land (Figure 4.3; for coefficient estimates see Appendix
B; Table B.4).
Table 4.5: Percentage of respondent households that were found to have complied with the five
community land use regulations. Only land inside the community agricultural areas was
considered.
Regulation
max 5 ha total area
max 1 ha paddy land
max 2 ha tree crop cultivation
spirit/burial forest
no buying/selling land
Andong Kraloeng
92
100
77
100
100
O Rona
56
94
81
100
92
77
52
all regulations
Figure 4.3 Predicted probabilities of compliance with IC regulations. The vertical dashed line
shows the average area claimed per household for the whole sample.
Increased household land holdings are associated with a significant reduction in the
probability of compliance with the set of five IC regulations. This effect is most
striking for non-IC members (Figure 4.3). It is also clear from comparing the curves
for non-IC and IC member households that once again there is a strong positive effect
on compliance associated with a household belonging to the village IC. Again, this is
expected given the two-tier system in place in O Rona, in which non-IC members
77
were afforded little right to land. Crucially, non-member households are not
represented on the committee that manages community land. It is unsurprising,
therefore, that households do not comply with regulations on which they have no say.
Interestingly, the correlation between compliance with IC rules and compliance with
the 2001 Land Law was negative. Comparing the curves in Figure 3 for households
with and without illegal land for both IC members and non-IC members separately, it
is evident that households with illegal land have a greater probability of being
compliant with the regulations governing land use in community areas over the full
range of total land holdings. This implies that households may seek to circumvent IC
regulations by clearing land outside of community areas, rather than breaking the
regulations within the titled area.
Perceptions of ICT
In addition to differences in land use, understanding and perceptions of the ICT
process differed significantly between the two villages. In Andong Kraloeng, 90% of
IC respondents displayed some understanding of the land use plan (this figure was
only 46% of IC respondents in O Rona). There was also a much greater level of
understanding in Andong Kraloeng that the responsibility for monitoring and
sanctioning households who were non-compliant with the land use plans was split
between the committee and the protected area authorities. Whilst 44% of IC
respondents in Andong Kraloeng mentioned the committee when asked who enforced
the land use plans, the figure in O Rona was only 17%.
With respect to how the land use plans were perceived, community members in
Andong Kraloeng were more likely to view the land use plans positively and to feel
secure about their land tenure. In O Rona, IC members frequently queried the validity
of IC regulations and restrictions on area, citing the failure of committee members (i.e.
those charged with managing community resources) and other community leaders to
observe them. The average land holdings for such leaders in O Rona was 8.6 ha, well
above the limit of 5 ha set down in IC rules. Conversely, community leaders in
Andong Kraloeng claimed an average of just 2.9 ha, which is below the village
average. IC members in O Rona also felt less secure regarding their land tenure. Only
6% of respondent reported that they felt secure in their tenure, with most respondents
78
worried about economic land concessions or powerful immigrants taking their land
(in comparison 72% of IC respondents in Andong Kraloeng felt secure).
Discussion
This research reveals households in two villages responding to land tenure reform and
drivers of land use change in different ways. In Andong Kraloeng, compliance with
both the legal land use plans that were created as part of the tenure reform process and
the community regulations that govern land use in community areas was relatively
high. In O Rona, a village that has been strongly exposed to market forces and
immigration, compliance rates were much lower, with extensive illegal land clearance
within conservation areas and illegal settlement of Khmer migrants observed. The
reasons behind these differences are complex and multi-faceted, yet they provide
valuable lessons for further implementation of ICT in Cambodia, and participatory
land use planning approaches more widely.
The most significant difference between the two villages is the degree to which they
have been exposed to outside interests. Milne (2013) reports that over 500 ‘outsiders’,
ranging from smallholder farmers to influential politicians, claim land inside the
provisional community area in O Rona. This is in addition to the families that have
moved into the village unopposed. As a direct result of losing land to these outside
interests, the boundaries of the O Rona ICT have been substantially redrawn from
those that were included in the original ICT application. That application contained
plans for 1477 ha to be granted for communal agriculture and residential use, while
only 648 ha has been included within the redrawn boundaries. The remaining 829 ha
has been lost to outside interests or non-IC households in the period between the
original application and receipt of ICT. Furthermore, the entire proposed western
NTFP forest area, which initially covered an area of 446 ha, has either already been
deforested or has been included as part of the 648 ha designated for community
agricultural use to compensate for the losses in this area.
It is debatable how much individual households have been involved in the sale of
community land to outsiders. Private access to land designated for indigenous
community title claims is often achieved through the use of intimidation,
misinformation and stealth (Fox et al., 2008; Milne, 2013). Under the 2001 Land Law,
79
communities are not granted disposal rights, which makes it illegal to sell land within
ICTs, but this has not stopped sales. Although little evidence of direct sales of land
parcels was found, a recent study of tenure policy changes introduced to O Rona after
the research presented here was completed reports frequent selling of plots by Bunong
to migrant Khmer families or outside land speculators (Milne, 2013). Such sales are
considered shameful and, hence, frequently carried out in private (Fox et al., 2008).
As such, reported compliance with the ban on buying and selling land in O Rona may
under-estimate actual sales. However, a comparison of land mapped for IC
households matched between the two time periods (and for whom the head of the
household has remained the same) shows that 74% of the land mapped in 2006 has
been retained. Given that households are expected to have given some of this land as
wedding gifts when their children married, it is possible that the selling of land that is
in use by IC households has been overstated (although this does not preclude the
selling of previously unclaimed community land).
Whilst the extent to which IC households have been involved in the sale of land to
outside interests is unclear, it is evident that the influx of external claims on
community land has had a negative impact on IC households. In interviews, IC
members frequently expressed frustration at their inability to exclude outsiders and
the loss of community land. For example, one woman said that “before people used to
try to stop the immigrants, but now we fear outsiders because they are rich and
powerful.” It is clear to them that their claim on the land, although recognised under
law, is insecure, thereby eroding one of the main benefits of the provision of tenure.
This in turn was used as a justification for clearing outside of designated community
areas. As one man put it, “we have to clear outside the community boundaries because
there is no more land available”. Despite these pressures, however, the ICT process
appears to have been partially successful in conservation terms, as IC members were
less likely to claim illegal land and when they did claim illegal land, they claimed
65% less land than non-IC members.
In Andong Kraloeng, the IC has made a concerted effort to prevent large scale
migration into the village, making access to land for outside interests significantly
more difficult (albeit the village is under less pressure than O Rona). In part, this can
be attributed to the village situation at the time land use planning was initiated. Unlike
80
in O Rona, where the two-tier system was necessitated in response to already high
levels of in-migration from non-indigenous households, Andong Kraloeng had very
few migrant households settle in the village prior to ICT. Of the 44 households
sampled in this research, only five were non-IC households. This has contributed to a
stronger sense of community identity and ownership of customary lands, as well as
greater belief in their right to exclude outside interests from their land. Consequently,
villagers report having turned away numerous migrant households since receiving
ICT. As such, those conditions most commonly associated with positive outcomes for
tenure reform have been achieved and compliance (particularly in regard to IC rules)
has benefited (Ostrom, 1990).
Despite the generally positive situation for Andong Kraloeng, there appears to have
been a breakdown of the traditional rotational cultivation system in response to the
emergence of cash crops over the past decade. This does not necessarily represent a
failing of the ICT process as no system can be expected to remain static over time,
particularly in the face of rapidly changing socio-economic conditions. It does,
however, underscore the importance of allowing for possible changes as part of the
planning process and of ensuring that the institutions that are created to manage
community areas have the capacity to be able to respond to those changes. As
agriculture becomes more commercialised in Andong Kraloeng, it remains to be seen
whether social institutions within the village can adapt, particularly if it brings
influential households into conflict with IC regulations. The results of the household
analyses support this, with those households with greater land holdings (i.e. older,
more affluent and more established households) more likely to be non-compliant with
community boundaries and regulations and also found to hold more illegal land if
non-compliant.
The results regarding the perceptions of IC members of the IC committees in each
village suggest that the social processes and institutions that support the management
of community land may play an important role in maintaining compliance with the
defined community boundaries and regulations. In Andong Kraloeng, committee
leadership is strong, with committee members seen to be compliant with community
regulations and legal boundaries, and ordinary community members display a greater
awareness of the committee’s role in managing community land. In O Rona, where
81
the overall perception the IC committee is poor, the ICT process is largely considered
to be an externally imposed intervention, and not supported.
This has important implications for how ICT and participatory land use planning
approaches are implemented elsewhere. In other villages in SPF, for example,
improved road access to more remote villages, and the increasing threat of land
alienation caused by economic land concessions, has led to the acceleration of the ICT
process. Whilst this has been forced by the changing conditions to which these
villages are exposed, the FA and WCS must be careful to ensure that the necessary
support is given to local social institutions, which are critical to the successful
implementation and sustainability of IC managed lands. With its origins in the
recognised need to support and empower local voices, this lesson is also particularly
relevant to the implementation of participatory land use planning approaches more
broadly.
Similarly important is the need to ensure on-going compliance monitoring, both
within community areas and the wider protected area, and to provide external support
for this where required. The considerable loss of community lands in O Rona, and the
perceived inability of IC members to prevent it, demonstrates that simply supporting a
community through the legal processes required to apply for tenure is not sufficient.
External political support may be necessary to assist communities in excluding
outside interests from illegal land grabs. Without this, the security that the
establishment of tenure should introduce may not materialise, leading to some of the
issues observed in O Rona.
It may also be necessary to provide stronger enforcement of community boundaries.
Unenforced boundaries will fail to induce sufficient incentive for local institutional
development and effective management of community resources. In Andong
Kraloeng, where the threat of external land grabbing is low and available land is far
from scarce, 26% of cultivated land has still been cleared illegally. While it is possible
that this is due to a lack of understanding of the rules that govern land use or of where
the boundaries of each area lie, it is more likely that these areas have been cleared
despite an understanding that to do so would be illegal. Effective policing of such
infractions is likely to increase the incentive for committees to manage land use
within community areas more effectively. This is supported by the result of the
82
community regulation compliance model, which suggested that households may offset
their non-compliance with community rules (to which they may feel more social
pressure to comply) through land use outside of community boundaries. As such,
protected area authorities should support community monitoring and exclusion of
outside interests within community boundaries, whilst ensuring adequate enforcement
of community expansion beyond those boundaries.
Conclusion
In this chapter, I have shown how two contrasting villages have coped with
implementing tenure reform. In one village, the evidence presented provides support
for the proposition that tenure reform in protected areas conducted with the
participation indigenous communities can be consistent with positive outcomes for
biodiversity conservation (in this case, retention of protected forest cover), whilst
providing communities with the rights to customary lands. In the other village, these
positive effects have been largely negated by severe disruption from outside interests,
powerful market forces and a failure in leadership on the part of those local
institutions created to manage communal lands. In such cases, it is in the interests of
both communities and those seeking to further conservation for local people to be
provided with the institutional support necessary to protect and manage their
resources effectively.
83
Chapter 5
5. Applying the carrot and stick in a Cambodian commons: an
experimental games approach to the investigation of
conservation incentives
Introduction
The institutional arrangements that govern individual behaviour can have substantial
consequences for the management of natural resources (Ostrom et al., 1990). In
situations where property rights are undefined or unenforced, extraction of resources
is limited by the ability of users to exploit them, their preferences to do so and the
institutions that govern extraction. Understanding behaviour under different
institutions is central, therefore, to the development of policies aimed at managing
resource extraction.
A common approach regarding open access environments is to predict extraction
based on the theory of rational self-interest, as characterised in Hardin’s ‘tragedy of
the commons’ (Hardin, 1968). Although this model accurately predicts behaviour in
many instances, it has been consistently shown to underestimate cooperative
behaviour under certain conditions (Gintis, 2000). Other studies have shown that
social norms, defined loosely as the standards or shared beliefs within a group
regarding how an individual ought to behave, can have a significant impact on
cooperative behaviour (Sethi & Somanathan, 1996; Ostrom, 2000; Biel & Thøgersen,
2007), with the extent to which norms affect behaviour heavily dependent on
individual preferences. Experiments investigating the cooperative behaviour of
individuals have found evidence to suggest that only a minority of subjects free ride,
with most being willing to cooperate if others reciprocate (Fischbacher et al., 2001;
Rustagi et al., 2010). These results suggest that, although some individuals conform to
the model of rational self-interest, they are unlikely to form the majority. Adding this
to the results of studies of social norms provides strong evidence to suggest that when
individuals are faced with a social dilemma there are many more determinants of
behaviour than assumed by the simplistic model of rational self-interest.
84
In addition to personal behavioural preferences, the institutional conditions under
which individuals make decisions are also a significant determinant of behaviour. For
example, a study of 15 ‘small-scale’ societies found that the behaviour measured
across different experimental games varied depending on the conditions present in
each setting (Henrich et al., 2005). Such experimental games are a common tool in
behavioural economics and have been applied to investigating the effects of different
institutions on behaviour, including resource extraction (e.g. Ostrom et al., 1994).
Games can be used to examine the role of socio-economic variables in decisionmaking and have the potential to enable the investigation of behaviour under
conditions that mimic the effects of different policies (Cardenas & Ostrom, 2004).
The results of such studies are varied and wide ranging but show that developing
appropriate institutions and governance systems is central to effective management of
resources (Vollan & Ostrom, 2010).
In this chapter, I present the results of a series of experimental games to investigate
how certain institutional arrangements affect extraction from a common pool resource
(CPR). I focus on the individual behaviour of smallholder farmers in a CPR setting
subject to combined institutional conditions. In so doing, I follow a similar
experimental design to that described in Travers et al. (2011), which enables the
comparison of the relative effects of different conservation interventions by
considering multiple treatments within the same experiment. These include external
enforcement and reward payment regimes similar to those found under the
mechanism of payments for environmental services (PES). Such payments are being
increasingly widely used in the context of protected area management, in conjunction
with more traditional enforcement activities, with conservation organisations
providing payments both collectively and directly to individuals in return for
engaging in, or desisting from, particular activities (for examples of how both
payment structures have been applied in northern Cambodia see Clements et al.,
2010).
The effect of enforcement of exogenous rules on behaviour has been the focus of a
number of experimental studies in the lab (e.g. Beckenkamp & Ostmann, 1999) and in
the field (e.g. Cardenas et al., 2000; Cardenas, 2004; Velez et al., 2006; Vollan, 2008;
Reichhuber et al., 2009), although the results are often contradictory. There is
85
evidence to suggest that under certain conditions existing pro-social behaviour can be
undermined ("crowded out") by rule enforcement, most notably when the probability
of detection is low, penalties for non-compliance are weak or intrinsic social cohesion
high (Ostmann, 1998; Cardenas et al., 2000; Vollan, 2008). Further evidence suggests
that even when cooperation is not reduced by enforcement, crowding out can occur
once enforcement has been removed (Reeson & Tisdell, 2008). Elsewhere, it has
been shown that exogenously imposed rules have the potential to reduce extraction
and increase the efficiency with which subjects negotiate game environments, even
when detection probabilities are low (Reichhuber et al., 2009; Travers et al., 2011).
This variation in the observed effect of exogenous regulation has also been found
within studies. For example in a study in rural Colombian villages, Velez et al. (2006)
found that the effect of exogenous resource controls varied significantly between
groups depending on the background of the participants. As such, it would appear that
while external enforcement institutions can be effective at reducing extraction from a
CPR, it is difficult to predict the outcome of such institutions prior to implementation.
In many conservation settings, those responsible for implementing rule enforcement
have little control over the size of penalties but may influence the probability that
non-compliance will be detected, for instance through greater investment in patrolling.
Hence, it is important to understand the effect that increasing the probability of
detecting rule breaking will have on compliance. In the lab, studies have shown that
subjects respond more to changes in the severity of sanctions than they do to the
probability of detection (Ostmann, 1998; Beckenkamp & Ostmann, 1999). Currently,
in the context of field based experimental games, there has been little work focused
on this question. This chapter seeks to address that omission by considering two
exogenous enforcement regimes: one mimicking the conditions often found in
developing country conservation settings in which the probability of detection is low
and one mimicking a situation in which conservation managers have invested in
enforcement and increased the probability of detection.
On the other side of the incentives scale, there is evidence that rewards can be highly
effective at inducing cooperation when the allocation of rewards is decided externally,
without cost to the recipients (Vollan, 2008; Travers et al., 2011; Narloch et al., 2012).
As with sanctions, there is a risk that exogenously awarded rewards may undermine
86
existing norms and serve to crowd out inherent cooperation (Frey & Jegen, 2001). In a
survey of attitudes regarding the site of a proposed noxious facility in Switzerland,
Frey and Oberhozer-Gee (1997) found that the offer of incentives as compensation
crowded out feelings of civic duty, resulting in lower levels of support. Conversely,
appropriation was reduced by an exogenous reward system in a CPR game amongst
pastoralists in Namibia and South Africa (Vollan, 2008). This effect was unchanged
even if a minority of the group voted in support of rewards when given the option of
voting for the different institutions within the game. In another study, Travers et al.
(2011) investigated collectively conditional rewards, whereby the reward was
conditional on aggregate group extraction falling within certain thresholds. It was
demonstrated that the level of conditionality (i.e. the group extraction threshold on
which rewards were conditional) was important in determining the effectiveness of
the rewards, with higher conditionality leading to more restraint in extraction. In
situations where participants have been allowed to choose between enforcement and
rewarding institutions, rewards have been found to be preferred (e.g. Vollan, 2008;
Sutter et al., 2010).
In this chapter, I examine the effect of two exogenous penalty and reward payment
regimes, as both stand alone and interacting interventions, on individual behaviour.
While the effect on extraction from a CPR of both enforcement and reward payment
regimes has been considered previously, here I investigate behaviour when both
conditions are present. This allows the study of potential interactions between the
disincentives created through enforcement and the positive incentives offered through
reward payments. This has particular relevance for conservation policy-making, as the
potentially conflicting interactions from different but simultaneously applied
interventions are rarely considered, despite their ubiquitous use in actual conservation
interventions (for example PES schemes within protected areas; Clements et al. 2010).
A common pool resource game
The CPR game used in this research was framed around the extraction of fish from a
communal pond (see Appendix C for the game script used). The harvesting of fish
was selected because fish is an important household resource and familiar to all
participants. Two other options were piloted (a generic, unitless resource and
bamboo) but feedback from the pilot suggested participants were better able to
87
understand the dilemma when it was presented in terms of harvesting fish.
Participants were split into groups of 10, (𝑛 = 10), and instructed that they would
make a series of decisions regarding the number of fish to harvest individually from a
communal pond. Each individual, 𝑗 ∈ {1, … , 𝑛}, could harvest 𝑥 fish to a maximum of
10, 𝑥 ∈ {0, … ,10}, each round, 𝑡 ∈ {1, … ,5}. In contrast to some other studies that
used non-linear payoff structures to model returns from extraction of a CPR (e.g.
Ostrom et al., 1992; Cardenas, 2000), a linear payoff structure was used because
initial piloting had shown that participants had difficulty understanding the
complexities of non-linear payoff structures. For each fish that an individual harvested
from the pond, they received 100 Riel with nobody else in the group benefitting. Each
fish left in the pond at the end of every round earned 12 Riel for all group members.
This was explained as the future benefit to the group of having fish in the pond. This
created a Pareto optimum at which no fish were harvested and a unique Nash
equilibrium at which every participant chose to harvest the full quota available to
them (10 fish). In total, 10 different experimental treatments were considered. For
certain treatments, participants could be subject to enforcement penalties for each fish
above a set threshold, e, with a probability of detection, λ, or reward payments, p (see
Table 5.1 for parameters used in each treatment). The individual payoff 𝜋𝑗 as
described by these rules is given by the following equation (see Appendix C for
specific payoff equations for each treatment):
𝜋𝑗 = 100𝑥𝑗 + 12(100 − ∑𝑛𝑖=1 𝑥𝑖 ) - λe + p
Eq. 5.1
Table 5.1: Treatment options for the CPR game. All treatments were played by 80 participants
with the exception of the private treatment, which was played by 360 participants. Treatment
parameters are given for the payoff equation (Equation 5.1).
Treatment
private
peer-pressure
weak enforcement
strong enforcement
individual payments
individual payments & weak enforcement
individual payments & strong enforcement
collective payments
collective payments & weak enforcement
collective payments & strong enforcement
Policy instrument
business as usual
law enforcement
law enforcement
PES
PES with law enforcement
PES with law enforcement
PES
PES with law enforcement
PES with law enforcement
88
e
125
125
125
125
125
125
λ
0.1
0.4
0.1
0.4
0.1
0.4
p
250
250
250
250
250
250
Private treatment
The private treatment was played in anonymous conditions, such that no individual
was aware of who else was in their group. After each round, participants were
privately told their own payoff, plus the total group extraction. Illiterate participants
were assisted in this. No communication between participants was allowed and
participants were made to sit apart. As such, this treatment provided a measure of the
internalised preferences of each participant in the absence of external controls on
behaviour.
Communication & peer-pressure treatment
In this treatment, participants were separated into two groups of 10 individuals.
Before individual decisions were taken, the group discussed the level of individual
extraction they thought acceptable, with any outcomes of the discussion non-binding.
They were then asked whether they had reached a group decision on the number of
fish that each person should take and, if so, that decision was recorded. Once
individual decisions had been made and the payoffs calculated, each participant read
out how much they had chosen to extract and their payoff. As such, the treatment
conditions allowed for individuals to experience shame, which has been shown to
increase pro-social behaviour in a social dilemma (Lopez et al.. 2012). Participants
were not permitted to lie to other group members at this stage. In this way, all
participants were made aware of the decisions and earnings of other group members.
This treatment served as a control for all further treatments, which were structured in
a similar manner in known groups of 10 individuals with group decision-making and
payoff reporting.
External enforcement treatments
In the external enforcement treatments, participants were subject to an imposed rule
enforced by an external agent. In this case, the rule was structured such that any
individual “caught” harvesting more than two fish per round would incur a penalty.
The threshold of two fish was set so that a small harvest was still seen as legitimate,
as past experience had shown that participants can struggle to understand the payoffs
if they are not allowed to harvest anything. This reflects conditions in many
conservation scenarios, in which only a certain level of appropriation is permissible.
89
The penalty was set at 125 Riel, 1.25 times the value of a harvested fish, for each fish
individuals harvested above the threshold of two. This was set to mimic the
enforcement regime for illegal harvest of natural resources in the study site, where
penalties are rarely enforced beyond confiscation of illegal harvests and tools. Once
individual decisions had been made and decisions and earnings read out, each
individual was asked to roll a ten-sided die, the outcome of which would determine
whether or not they were monitored that round. If they were monitored and were
found to have harvested more than the threshold, then they received the proportional
penalty of 125 Riel per fish. Two treatments were used: for the weak enforcement
treatment, the probability of being monitored was set at 0.1 and for the stronger
treatment the probability of being monitored was four times greater at 0.4. For
example, if an individual chose to harvest four fish during the weak enforcement
treatments and rolled a 2 on the die, they would not be monitored and their noncompliance would go unpunished. If they behaved in the same way during the strong
enforcement treatments, they would be monitored and would be fined 250 Riel (125
Riel for each fish harvested above the threshold of two).
Individual payments treatments
For the individual payments treatments, participants were offered reward payments if
they kept their harvest to two fish or fewer. Compliant participants were eligible to
receive a bonus payment of 250 Riel, equivalent to 2.5 times the value of a single fish.
The bonus was set in this way so that the individual opportunity cost of not harvesting
at or close to the Nash equilibrium was not met by the bonus. This was intended to
mimic the level of potential incentive payments in the study site, which are unlikely to
cover the full opportunity cost incurred by foregoing extraction. In this case, it was
assumed that monitoring would be perfect. For the two treatments that combined the
individual payments treatment with the two external enforcement treatments,
participants could receive a payment for keeping extraction below the threshold but
also faced the possibility of a penalty for over-extraction.
Collective payments treatment
In the collective payments treatments, participants were again offered incentive
payments but this time the payments were conditional on aggregate group behaviour.
90
If the total group extraction was equal to or less than 20 fish, every individual would
receive a bonus payment of 250 Riel. If the total harvest was greater than this
threshold, none of the participants would receive a payment, irrespective of their
individual harvest. This treatment modelled incentive schemes where participants are
contracted on a collective basis rather than individually. Again, it was assumed that
monitoring was perfect. For the two treatments that combined the collective payments
treatment with the two external enforcement treatments, participants could receive a
payment if the group kept extraction below the threshold but also faced the possibility
of a penalty for over-extraction.
Theoretical prediction and hypotheses
In this section, the behaviour predicted by the standard model of rational self-interest
is presented, as well as a priori hypotheses regarding the expected behaviour under
different treatments of the CPR game that follow from the background literature
presented in Section 5.1.
Standard theory prediction
Under the standard Homo economicus model of rational, selfish behaviour,
participants of the CPR game described in Section 5.2 should remain unaffected by
the different treatments. For the control peer-pressure and private treatments, in which
there were no reward payments or penalties, the Nash Equilibrium is for all
individuals to extract 10 fish, the maximum possible. For the enforcement treatments,
the marginal expected penalty for extraction is lower than the marginal return (even
under the higher probability of detection), and therefore does not alter predicted
behaviour. Likewise, the theoretical prediction for the payment treatments remains
unchanged, as the payment received for extracting below the threshold is lower than
the individual return of extracting the full allowance.
Hypotheses
H5.1: probability of detection in rule enforcement
H5.1.1 As the probability of detection is increased, the expected penalty for noncompliance increases also. As discussed above, under institutional conditions that
allow for the enforcement of rules with penalties, the behaviour of risk neutral
91
individuals conforming to the model of rational self-interest would be expected to be
unaffected by changes to the probability of detection, provided that the expected
penalty remained less than the marginal return of non-compliance. As such, for the
experimental conditions described above, it is expected that, under enforcement
conditions only, there will be no significant difference in extraction from the pond
between the two enforcement treatments.
H5.1.2 Previous studies have found evidence to suggest that crowding out of intrinsic
motivation is likely to occur when weak exogenous rule enforcement conditions are
imposed (e.g. Cardenas et al., 2000). However, only weak evidence of such an effect
was found by Travers et al. (2011) under similar conditions as those studied here.
Therefore, it is expected that no crowding out will be observed under enforcement
conditions only, compared to the extraction based on intrinsic motivations observed in
the control.
H5.2: individual vs collective payment structure
H5.2.1 Both payment structures tested are expected to result in significant reductions
in extraction in comparison with that observed for the control treatment. Given that
the experimental design of this research closely matches that of Travers et al. (2011),
which found no significant difference between two individual and collective payment
structures in the effect on extraction from a common pool resource, the individual and
collective treatments are expected to perform similarly in reducing extraction.
H5.2.2 It is also predicted, following the results of Travers et al. (2011), that the
collective payment treatment will have secondary benefits, such as carry over effects
into subsequent treatments, which are not observed under the individual payments
treatment.
H5.3: combined payments and penalties
It has been shown that the phenomenon of crowding out is more likely in situations in
which pro-social behaviour would otherwise be strongest (Vollan, 2008). As such, it
is expected that combining the payment-based treatments, which are expected to
promote pro-social behaviour, with the two enforcement institutions will result in
92
crowding out and increased levels of extraction in comparison with the payment
structures operating in isolation.
Methodology
The CPR game was conducted in 13 villages between March and April 2012. Each
village was categorised a priori into one of three village types using two key variables
for guidance: distance to the nearest all-day market and the proportion of Bunong
households in the village (for values of selection variables for each village see
Appendix C; Table C.1). These variables are thought to influence a range of village
characteristics, including poverty, livelihood strategies and village cohesion. Access
to each village plays an important role in governing livelihood decisions as it has a
strong effect on the sale price of different commodities and the cost of transporting
goods to market. In reality, access is a combination of distance and the quality of road,
but road quality is particularly difficult to measure as it varies annually and
throughout each year. While many of Cambodia’s minority indigenous peoples have
become integrated into Khmer society, often to the point where they identify
themselves as Khmer, the Bunong have maintained many of their customary practices,
including their own language. The proportion of Bunong households in a village,
therefore, provides a useful proxy measure of the level of integration with Khmer
society.
For each village type, 120 villagers were selected to participate in the games.
Households were selected at random from a list kept by the village chief. The head of
each household was then contacted by the village chief and invited to participate in
the games. If they were unavailable or chose not to play, another adult was invited
from the same household. It was made clear to all selected households that they were
not required to participate, and a member of the research team was present to ensure
that no household was coerced. A reserve list of randomly selected households was
produced to ensure that 20 individuals agreed to play each session in case households
were unavailable or chose not to participate. As there were fewer than six villages in
two of the three village types, the number of participants was higher for some villages
than for others. Sample sizes were approximately proportional to population size.
Participants were all aged over 18 and where possible only one person per household
participated. In certain cases, this was not possible due to the small size of some of the
93
study villages and, in those cases, members of the same household played on the same
day but never interacted in the game. The experiments were held in the local school
for the majority of study villages. In some villages, however, there was no school or
the school was not available. In such cases, appropriate houses were identified with
sufficient space for participants to sit in groups or separate for the private treatment.
The CPR game followed a similar experimental design to that described in Travers et
al. (2011). It was conducted over 18 separate days, with 360 participants (20 per day).
Each participant played three different treatments: the private treatment plus two
others. The order in which the treatments were played was changed for each session
to control for possible ordering and earnings effects. To avoid individuals’ decisions
being governed by repeatedly playing with the same people, group composition was
controlled to minimise the number of participants who played each treatment in the
same group. All treatments were repeated for a total of five rounds. Payments for all
rounds were made after the game had finished to avoid payoff effects influencing
behaviour. Each individual was given a payoff sheet on which the payoffs earned in
previous treatments were written and was paid privately, with the majority of earnings
totalling between four and six US dollars. Daily farm wages in the study area are
usually five dollars so the game payments were roughly equivalent to paid labour. All
sections of the experiment were run by a team of four: the lead author, two Khmer
research assistants and a Bunong translator to assist participants with lower Khmer
skills.
After an explanation of the game, two practice rounds were played following the same
format as the private treatment. Participants marked decisions by circling the number
of fish they wanted to harvest (see Appendix C for individual decision sheet).
Individual earnings and total group extraction were written on all sheets after each
round. Once the practice rounds were completed, participants were asked five
multiple-choice questions to test their understanding of the main principles of the
dilemma. On completion of this test, the answers were explained to help those
participants that still had problems with understanding. Individual scores from these
tests were checked in the regression analysis described below but were found not to
be significantly correlated with game behaviour. During the experiment, the private
treatment was played in a single room and all other treatments were played in separate
94
rooms such that it was not possible for different groups to hear what was happening in
groups other than their own.
Results
Generalised linear mixed effects models (GLMMs; Zuur et al., 2009) were
constructed to analyse the results of the game2. The logit link function was used due
to the binomial error structure of the number of fish taken by an individual, with
Laplace approximation used to estimate the model parameters (Bolker et al, 2009).
Model selection for fixed effects was carried out by comparing Akaike information
criterion values (AIC; Akaike, 1974). The final model was selected using step-wise
selection and comparing ∆AIC values. Following Burnham and Anderson (2002), if
the difference in AIC values between models was less than two, the most
parsimonious model was selected. If the ∆AIC values were greater than two, the
model with the lower AIC value was selected. The random effect structure, including
variables for village, day number, group and individual, was investigated using
likelihood ratio tests (Bolker et al, 2009). Once both the random and fixed effect
structures had been selected, visual validation was conducted to check for residual
normality, heteroskedasticity and for possible correlations between fixed effects and
the residuals. Each model was also checked for over-dispersion by comparing the sum
of squared Pearson residuals with the approximate residual degrees of freedom.
The results of the CPR game are best compared against the baseline created by the
control peer-pressure treatment. In this treatment, the mean individual extraction was
4.9 fish per round. This closely matches the results of Travers et al. (2011). As
expected, the majority of treatments performed better than the control at reducing
extraction from the CPR (Table 5.2; Appendix C, Figures C1-C3). Whilst some
treatments achieved an equilibrium state with average harvest remaining effectively
constant over the five rounds, extraction in other treatments had not reached
equilibrium at the end of the game.
2
All models were analysed using R version 2.12.0 (R Core Development, 2010) in
the open source software package RStudio version 0.96.228 (RStudio, 2012). Within
R, the lme4 package version 0.999375-36 (Bates & Maechler, 2012) was used to code
the GLMMs.
95
Table 5.2: Descriptive statistics for key explanatory variables. Sample standard errors are shown
in parentheses.
Variable
male participants
Bunong participants
mean age
mean years in education
mean number of fish taken during each treatment:
practice rounds
private
peer-pressure
weak enforcement
strong enforcement
individual payments
individual payments & weak enforcement
individual payments & strong enforcement
collective payments
collective payments & weak enforcement
collective payments & strong enforcement
discussion periods ending in decision
Statistic
45%
62.9%
37.5 (13.4)
2.6 (3.2)
4.5 (2.1) fish
5.5 (3.0) fish
4.9 (2.8) fish
4.5 (2.9) fish
5.9 (2.8) fish
2.2 (1.9) fish
3.8 (2.8) fish
3.1 (2.7) fish
2.1 (2.2) fish
2.9 (2.4) fish
2.6 (1.6) fish
97.1%
The effect of treatment and socio-demographics on individual extraction
The model for the individual number of fish harvested each round shows a mixed
picture, with the private and both enforcement treatments having no significant effect
on individual extraction compared to the peer pressure control treatment (Table 5.3,
column 2; difference in fish = -0.1 and 0.4 for the weak and strong enforcement
treatments respectively). This matches the prediction made under hypothesis H5.1.1
that increasing the probability of detection, without the expected penalty being higher
than the marginal gain in harvesting extra fish, would not result in a significant
change in extraction behaviour. Conversely, all treatments that included individual or
collective payments resulted in reductions in extraction. For the two payment
structures in the absence of enforcement, there was no significant difference between
their effects on extraction; with the mean reduction in individual extraction estimated
to be 2.6 fish (SE = ±0.4) for individual payments and 2.2 fish (SE = ±0.4) for the
collective payments (Wald test, P = 0.447). This supports the prediction made in
hypothesis H5.2.1 that no significant difference would be observed in extraction
under the two payment structures.
96
Table 5.3: Parameter estimates for the selected CPR game generalised linear mixed models, with
the number of fish harvested by an individual in a single round as the response variable. The
difference in the number of fish taken was found by calculating the expected number of fish
taken for each variable and comparing this with the number expected for the intercept. Standard
errors in terms of number of fish are shown in parentheses. A comparison of the effect of each
treatment for round 1 and round 4 is given in Supplementary Materials Table C.4. The full list of
variables modelled, plus their explanations, are given in Supplementary Materials Table C.2.
Significance values: ns = not significant; (*) = P < 0.1; * = P < 0.05; ** = P < 0.01; *** = P < 0.001.
Reference level treatment indicated by ref.
Explanatory variables
intercept
treatment:
peer pressure
private
weak enforcement
strong enforcement
individual payments
collective payments
individual payments &
weak enforcement
individual payments &
strong enforcement
collective payments &
weak enforcement
collective payments &
strong enforcement
education
decision
practice
round:
round 2
round 3
round 4
round 5
random effect:
individual
group
session
Peer pressure treatment
as ref
Individual
payments & strong
enforcement as ref
Estimate
P
Estimate
P
3.938 (0.724)
*
2.815 (0.634)
**
ref
-0.105 (0.643)
-0.097 (0.838)
+0.383 (0.852)
-2.567 (0.434)
-2.243 (0.514)
ns
ns
ns
***
***
+1.123 (0.806)
+1.018 (0.641)
+1.026 (0.839)
+1.506 (0.810)
-1.444 (0.457)
-1.121 (0.549)
ns
(*)
*
ns
**
(*)
-1.927 (0.575)
***
-0.804 (0.608)
ns
-1.123 (0.709)
(*)
ref
-
-1.774 (0.641)
**
-0.651 (0.602)
ns
-2.252 (0.534)
***
-1.129 (0.508)
*
+0.056 (0.022)
-0.750 (0.185)
+0.191 (0.038)
**
***
***
+0.048 (0.019)
-0.614 (0.147)
+0.163 (0.033)
**
***
***
+0.855 (0.081)
+0.577 (0.080)
+1.009 (0.081)
+0.835 (0.081)
***
***
***
***
+0.755 (0.074)
+0.503 (0.072)
+0.898 (0.076)
+0.736 (0.075)
***
***
***
***
SD = 0.484
SD = 0.561
SD = 0.591
SD = 0.484
SD = 0.561
SD = 0.591
The picture becomes more complicated when the treatments integrating payments
with enforcement regimes are considered. For the treatments with individual
payments, enforcement appeared to undermine the incentives for extraction reduction,
with both levels of enforcement producing a weaker effect on extraction than the
payments on their own. This crowding out was more severe for the stronger
97
enforcement regime (1.1 fewer fish extracted than in the control, compared to 2.6
fewer for individual payments alone) than the weaker one (1.9 fewer fish). For the
collective payment treatments, no undermining effect was observed for either of the
two enforcement regimes (1.8 fewer fish than the control with weak enforcement and
2.3 fewer fish with strong enforcement, rather than 2.2 fewer with just collective
payments).
Running the model again with the individual payment with strong enforcement as the
reference level provides a clearer picture of the relative performance of the different
payment treatments (Table 5.3, column 3). This shows that adding strong enforcement
to individual payments significantly increased individual extraction relative to the
individual payments treatment (1.4 fewer fish under individual payments alone than
payments with strong enforcement), collective payments treatment (1.1 fewer fish)
and collective payments with strong enforcement treatment (1.1 fewer fish). On this
basis, it appears that although there was no difference between the performance of
individual and collective payments under zero enforcement, performance for
individual payments was crowded out slightly under weak enforcement and
significantly under stronger enforcement regimes. Under no combination was there
any meaningful increase in compliance under integrated treatments relative to the two
payment treatments. This matches the effect expected under hypothesis H5.3, with
exogenous enforcement crowding out the strong pro-social behaviour observed when
payments were offered with no enforcement. However, the lack of an effect when rule
enforcement was added to collective payments is contrary to predictions and suggests
that further thought is required to understand why crowding out occurs under some
conditions and not others.
In addition to the treatment played, a number of individual demographic and
livelihood variables were tested, of which only one was included in the final selected
model. This was the number of years of education that each participant had received.
Despite the fact that a logistic regression was used in the analysis, the effect of
education was linear over the range of years spent in education observed for the
sample. As such, each additional year in education was found to result in a small
increase in the number of fish taken (diff. in fish = 0.06). None of the other variables
98
tested were included in the final model, meaning that no effect was found for
participant age, gender, ethnicity or any of the livelihood variables considered.
Two other variables were found to have a significant effect on individual extraction.
Group decision-making (i.e. when a group made a collective decision during the
discussion period) resulted in a reduction of 0.8 fish per group member when a
decision was recorded, mirroring a similar result to that found in Travers et al. (2011).
This shows the strong impact that collective decision-making can have on individual
harvesting decisions. Finally, the mean individual extraction from the two practice
rounds was also found to be significantly and approximately linearly correlated with
extraction during the game itself. This suggests that behaviour within the game was
governed not only by the conditions of each treatment but also by inherent
preferences. Participants, who took more than others during the practice, also took
more in the other treatments they played, irrespective of the difference in conditions.
Three random effects were also included in the selected model: individual, group and
session. The variation explained by each random effect was roughly comparable
(Table 5.3). This indicates the importance not just of individual decision-making but
also of group interaction, with both the groups in which participants played individual
treatments and the group of 20 participants selected to play the CPR game for a single
day session included in the model. This matches a similar result in Hayo and Vollan
(2012), who found that adding a group dummy variable to their analysis of CPR
extraction in Southern African rangelands improved the explanatory power of their
model.
One variable that was found to be important by Travers et al. (2011), but was not
included in the selected model here, was the previous treatment played. Previously, it
was found that individual extraction was significantly reduced after participants had
played treatments that promoted a degree of self-organisation within each group.
Under the conditions presented by the set of treatments considered here, this would
suggest that the three collective payment treatments would reduce extraction in
subsequent treatments played. The difference between the AIC of a model containing
a variable for the previous treatment played and the selected model was 6.3, which
suggests that the previous treatment played has no effect on extraction. A closer look
at the parameter estimates of each individual treatment confirms that only one
99
treatment had any meaningful effect on extraction in future treatments; the simple
collective payments treatment (diff. in fish = -2.1, p = 0.004). This is comparable with
the effect size found in the previous study, and supports hypothesis H5.2.2 that
although performance under the two payment structures was not significantly
different, collective payments offer secondary benefits that are not observed for
individual payments. The fact that no effect was recorded for the combined collective
treatments suggests that, although overall performance was not reduced, integrating
enforcement with an payment-based intervention can crowd out at least one of the
positive effects of self-organising.
Discussion
This chapter builds upon existing research that investigates how different institutional
arrangements affect individual extraction behaviour in a CPR scenario. By
investigating behaviour under more complex and realistic institutional conditions, in
which both payments and penalties are applied simultaneously and separately, it has
been possible to examine how individuals respond to such conditions, particularly
whether penalties can crowd out the pro-social behaviour observed under reward
payment treatments.
The fact that little difference was found in performance when the two payment types
(collective and individual) were considered in isolation may have implications for the
structuring of payments in the study site and more widely. Collective payments are
thought to offer potential advantages over individual contracts, such as reduced
transaction costs (Wunder, 2007), and to encourage cooperation through peer
monitoring and peer pressure to comply. Structuring payments collectively can also
generate secondary benefits because of the institutions created to manage them
(Clements et al., 2010). Collective payments are, however, potentially less appealing
to recipients where social cohesion is low, as the ability to control the extraction of
others in their group is likely to be weak. Should individuals within the group ignore
the incentive, their behaviour has the potential to impact upon compliant individuals.
This may in turn lead to the non-compliance of individuals who would have
cooperated if contracted individually, as they are faced with incurring the opportunity
cost of foregoing extraction with little hope of receiving the compensatory payment.
100
The more simplistic individual payments are less vulnerable to the concerns raised for
collective payments and, consequently, offer the potential for improved performance
in situations where social cohesion is low. This reflects the results of Narloch et al.
(2012), who found that individual payments improved contributions in a public goods
game but that collective payments were ineffective. In that study, groups were unable
to make agreements or discuss behaviour, which is likely to limit social cohesion.
However, the set-up in this research led to similar results from both individual and
collective payments, although the transactions costs of setting up individual incentive
agreements are likely to be higher than those for collective payments in the Eastern
Plains Landscape. A preference for collective over individual payments in this system
is further supported by the carry-over effects observed for collective payments. The
lower extraction recorded for treatments following collective payments suggests that
participants were able to maintain the same level of cooperation even after the offer of
a payment had been removed. Although the mechanism for this carry over effect is
currently unclear, Travers et al. (2011) observed the same effect in treatments thought
to promote self-organisation amongst game participants. This supports the notion that
structuring payments collectively has the potential to empower recipients and, hence,
may lead to a more sustainable solution than if payments were simply contracted on
an individual basis.
The role of enforcement and crowding out
Neither of the two external enforcement treatments achieved a significant reduction in
extraction over the control. This follows theoretical predictions, as the incentives in
both treatments were insufficient to move the Nash Equilibrium away from each
individual extracting the maximum possible. Both treatments resulted in lower social
efficiencies than found for the control, with the stronger treatment lowest due to the
higher penalty rate. Even though participants would have been aware that these
treatments were resulting in lower individual payoffs, the penalties were insufficient
to reduce extraction. This corresponds to the findings of a number of other studies (e.g.
Ostmann, 1998; Cardenas et al., 2000). There was, however, no evidence that the
enforcement treatments in isolation led to crowding out of the baseline cooperation
measured in the control, despite the generally strong social cohesion of the subject
101
pool. This is at apparent odds with the findings of Vollan (2008) that crowding out
occurs when people have strong social cohesion and are subject to external penalties.
The effect of enforcement on the cooperation achieved under the two payment
structures is more informative. It is rare for incentive schemes in conservation to
operate in isolation, without some form of enforcement (often imposed by an external
agent). As such, it is important to understand how these different policy instruments
interact. None of the treatments that combined enforcement with payments improved
on the performance of payments alone. In many ways, this is unsurprising as both the
collective and individual payment treatments lowered mean extraction to
approximately two fish per round for each individual, the same as the threshold for
receiving a payment. Given that the enforcement regimes used the same threshold,
improvements in performance would have been expected to be restricted only to those
individuals who took more than the threshold.
In the case of the weaker enforcement regime, adding complexity to the institutional
conditions experienced by the participants appears to have weakened the response to
the incentives created by the two payment structures, as both saw a slight decline in
their effectiveness. This is an example of how externally imposed rules have the
potential to crowd out cooperation achieved under other conditions. Other authors (e.g.
Cardenas et al., 2000) have suggested that this crowding out effect is caused by
externally imposed rules promoting individualistic behaviour, thereby undermining
cooperative behaviour. This was supported by the way in which participants
responded to the enforcement mechanism during the game, with several participants
commenting during the discussion periods that extraction decisions should be up to
the individual because of the risk of being penalised.
For the individual payment treatment, the crowding out effect was strongest when
combined with the stronger enforcement regime. In some sense this is counter to
expectations, as increasing the probability of detection reduces the expected payoff of
breaking the rule limiting extraction. One possible explanation for this result is that
the high probability of monitoring acted to highlight the importance of the
enforcement mechanism in participants’ minds and, hence, further promote
individualistic behaviour. Participants understood that group members who took more
than the threshold of two fish did so at their own risk, thus reducing the incentive for
102
group decision-making and individual compliance. The response to the strong
enforcement regime when coupled with collective payments was very different. In
this case, several participants were observed using the presence of penalties in the
treatment as a means to encourage non-compliers to cooperate. By so doing, they
were using an externally imposed institution to promote behaviour in the group’s best
interest, an action for which there was little incentive in the case of the individual
payments.
This apparent difference in how the two payment structures interacted with the
enforcement regimes demonstrates the potential for individuals to respond very
differently to seemingly similar sets of incentives. In the case of individual payments,
increasing the probability of rule breaking detection strengthened the crowding out
effect, and increased extraction, but the same mechanism served to strengthen group
decision-making for collective payments. This finding builds on the work of Vollan
(2008), as it shows that crowding out can occur when both supportive and controlling
institutions are present, but only under certain conditions.
Policy implications
Whilst many previous studies have investigated the effect of both positive and
negative sets of incentives on extraction behaviour in the commons, the results
presented here have built up a more complex picture by comparing the effects of
multiple simultaneous interventions, including several that integrate seemingly
opposing sets of incentives. These results can help guide policy decision-making.
One of the principal dilemmas facing conservation organisations is how to provide
incentives to change illegal behaviour. Typically, this has been attempted through the
implementation of various enforcement measures but there are a number of barriers to
successful implementation of such measures. These include, amongst others, weak
local institutions (Barrett et al., 2001), poor support for natural resource laws from
national governments and logistical and financial constraints. More positive measures
can be employed but this raises questions about the value and form of possible
incentives. Some authors have argued that it is necessary to compensate the full
opportunity costs of foregoing the illegal behaviour but this approach has clear issues
associated with it. In particular, the opportunity costs of many illegal activities, such
103
as the clearance of forest for commercial agriculture, may be prohibitively high. The
results of the payment treatments are encouraging in this respect, as all treatments,
including those integrated with enforcement regimes, resulted in significant
reductions in individual extraction, and yet only covered 31% of the individual
opportunity cost of not harvesting the full 10 fish. This suggests that, contrary to the
model of selfish rationality, individuals will alter their behaviour even when the pure
financial incentives are insufficient to suggest that they would do so.
The results of the integrated payment and enforcement treatments are of greater policy
relevance. These demonstrate that focussing solely on the effects of positive
incentives may overestimate the impact of such policies by failing to take account of
the possible crowding out effect of also providing negative incentives by externally
imposed enforcement of rules. There is also evidence here to suggest that these risks
can be minimised by adopting a collective payment structure rather than an individual
structure. This would also offer the advantage of promoting a greater degree of selforganisation amongst resource users and lowering transaction costs. In the context of
payments for environmental services, collective payments may therefore provide a
more effective solution for encouraging reductions in individual resource use.
Finally, in this chapter, I have considered a series of exogenous conservation
interventions that might be employed to encourage pro-conservation behaviour in the
study area. In the context of Seima Protection Forest, the results presented suggest
that certain policy options could be used to reduce extraction from the reserve. Both
the individual and collective payment treatments brought considerable reductions in
comparison to the baseline behaviour. Conversely, increasing investment in patrolling
may not bring about any significant benefits or could even be counter-productive if
combined with individual payment schemes designed to encourage resource
conservation. Consequently, it is recommended that careful consideration be given to
possible unwanted effects, such as crowding out, during the design phase of future
payments schemes proposed for the site.
104
Chapter 6
6. Investigating incentive based approaches to biodiversity
conservation through scenarios
Introduction
It has long been understood that local communities shoulder a disproportionate share
of the costs of biodiversity conservation, whilst benefits accrue largely at national or
global scales (Wells, 1992; Balmford & Whitten, 2003). The establishment of
protected areas (PAs), for example, is often driven by values surrounding the
importance of endangered species or habitats, which may not be shared by those most
affected by conservation activities (Roe & Elliott, 2006). Conversely, PAs have often
been associated with displacement, food insecurity and the loss of livelihoods
(Ghimire & Pimbert 1997; Adams et al., 2004; Colchester, 2004; Brockington & Igoe,
2006; Cernea & Schmidt-Soltau, 2006), with significant negative impacts for local
people. Recognition of this inequality and the resentment it can create has helped to
give rise to a raft of incentive based approaches to conservation. Notable examples
include integrated conservation and developments projects (ICDPs; Wells & Brandon,
1993; Barrett & Arcese, 1995), community based natural resource management
(CBNRM; Kellert et al., 2000; Singleton, 2000) and payments for ecosystem services
(PES; Ferraro & Kiss, 2002; Wunder, 2007; Engel et al., 2008). While the motivation
behind these different approaches may vary, the principle that local people are
incentivised to cooperate with conservation efforts is common to all. However,
despite the long period of interest in incentives and their increasing use in
conservation, questions remain about how best they might be structured. For instance,
should incentives be collective or targeted at individuals? Are direct payments
preferable to in-kind support? How do the effects of using payments compare with
those of alternative top-down approaches, such as law enforcement? These are all
questions that should be answered prior to the implementation of conservation
policies, and yet their answers are not clear-cut.
For contexts in which the use of resources is restricted or illegal, such as within
protected areas, the appropriateness of using incentives as a means to encourage pro-
105
conservation behaviour has been questioned. Concerns have been raised regarding
both the additionality and validity of providing incentives to discourage a behaviour
that is already illegal (Börner & Wunder, 2008). However, where resources are
lacking for effective protected area management, incentive schemes may still be a
legitimate means of changing behaviour (Clements, 2010; Petheram & Campbell,
2010). In such cases, it may not be necessary or appropriate to cover full opportunity
costs, as incentives can augment existing preferences to obey the law. In this context,
it has been suggested that in-kind support, such as the setting up of a community fund
for infrastructural development or livelihood enhancement (e.g. agri-environmental or
alternative livelihood schemes) to increase the profitability of existing land uses, is
preferable to direct payments (Wunder, 2005; Asquith et al., 2008). There is a
corresponding risk, however, that such an approach could weaken the conditionality
attached to the provision of incentives, as such support could be difficult to withdraw
(Sommerville et al., 2010), and would also vary in the extent to which different
resource users benefit (Lee & Mahanty, 2009).
Assessing the future response of those targeted by conservation programmes is not an
easy task, particularly for contexts, such as those commonly found in developing
countries, in which institutions are weak (Barrett et al., 2001) and property rights illdefined (Geist & Lambin, 2002). Consequently, decision-making under different
economic incentives is often conceptualised in terms of maximising utility (Barrett &
Arcese, 1998; Damania et al. 2005). However, this approach is problematic. Whilst
decision-makers are commonly modelled as rational actors, who seek only to
maximise their own benefit, empirical evidence suggests that most decision-making is
influenced by other-regarding preferences (Gintis, 2000). For environments where
property rights are uncertain, such as for many common property resources,
predictions of behaviour become particularly complex. Similarly, the opportunity
costs borne through cooperating with conservation initiatives not only incorporate
tangible financial costs, such as those of lost earnings through restricted access of
resources, but may also include other elements (such as foregone recognition of
tenure), which are harder to quantify. Issues such as these, coupled with a paucity of
data relating to potentially unknown factors that influence decision-making, can make
predicting behaviour more challenging. Similarly, testing approaches in real life is
106
also difficult, as pilots can be prohibitively expensive and experimental approaches,
such as randomised controlled trials, even more so (Agrawal, 2014).
One way to overcome the difficulties in predicting behaviour is to turn to methods,
such as scenario-based interviewing, which allow for the investigation of decisionmaking under a range of credible futures prior to the implementation of an
intervention. While predicting future behaviour is problematic, placing limits on the
length of time and changes considered can act to reduce the complexity and produce
meaningful results (Gordon, 1992). Scenarios presented in the form of qualitative
narratives (e.g. Cinner, 2009) have the advantage that they can be easily understood
and, therefore, allow for the exploratory examination of behaviour in contexts where
respondents may struggle to place values on complex hypothetical situations (as in
contingent valuation). Discussing the future in this way provides valuable insight into
not only how people are likely to respond to the scenarios presented, but also the
reasons why they might respond that way.
Such approaches may also serve to help minimise the risks presented by heterogeneity
amongst the target populations for conservation policies, or by exogenous changes
such as external market fluctuations. Accounting for such complexity is one of the
principal challenges facing conservation; something that integrated conservation and
development projects are often said to have failed to do (McShane & Newby, 2004;
Blom et al., 2010; Waylen et al., 2012). In part, this is because conservation often
operates within highly complex socio-ecological systems in which relationships
between society and natural systems are dynamic and multi-scale (Berkes, 2004).
Even at the site level, heterogeneity within target populations may be high (Chan et
al., 2007; Waylen et al., 2013). Scenario approaches enable the response to
conservation policies by different agents to be tested and, hence, the extent of
homogeneity of response to be estimated for a target group.
In this chapter, scenario interviews are used to examine the potential outcomes of
different approaches to changing incentives within Seima Protection Forest. This is
done through an analysis of the stated responses of smallholder farmers from several
villages within the project area to seven future scenarios, which include exogenous
changes to the sale price of cassava (the dominant agricultural commodity in the area)
and different intervention options aimed at reducing deforestation at the site
107
(increased enforcement effort, communal and individual conditional payments and a
village fund for infrastructural development). For each scenario, interview
respondents were asked how their land use and livelihood practices might change and,
hence, the responses given provide an indication of the expected variation in farmer
reactions to the intervention options under consideration for the study site. Within this
methodological framework, the predicted effectiveness of each option for
incentivising pro-conservation behaviour (in this case, reduced forest clearance) is
investigated and compared against the response to exogenous changes in the price of
cassava. By analysing responses at the household level but within three distinct
livelihood zones, the effect of economic well-being, livelihood strategy and sociodemographic variables on the responses given to each scenario are examined. In this
way, I seek to identify whether, in the context of this case study, opportunities exist to
target interventions towards those households or livelihood zones most likely to
respond positively.
The Seima Protection Forest REDD+ Demonstration Project
Proposals for the REDD+ project at Seima Protection Forest are well advanced and
focus on the core protection zone, with 20 participating villages located within or in
close proximity to this zone. Household livelihoods within the project area are based
on subsistence agriculture and the collection of forest products (Evans et al., 2003).
Increasingly, however, farmers are turning towards the production of cash crops, such
as cassava and cashew, and encroaching further into the protected area (Chapter 4).
Immigration of households seeking land from neighbouring provinces has helped to
drive expansion of agricultural land and fuel encroachment (Chapter 3; Chapter 4;
Milne, 2013). Whilst these two processes are helping to drive deforestation within the
proposed REDD+ project area, the rate and extent of change vary between villages.
A programme of participatory tenure reform is currently underway or complete in 15
of the project villages (A. Diment, pers. comm.). Under this scheme, communal land
title is granted to each officially recognised indigenous community, with land inside
village boundaries divided into different use areas. Land use inside the communal title
is governed by a set of rules designed to protect customary practices and ensure longterm sustainable use of communal land. Clearance of forest is only permitted inside
designated community areas. Hence, clearance outside the titled area is illegal and
108
subject to enforcement by PA authorities, whilst clearance inside community areas is
governed by village institutions. Clearance rates are high both inside and outside titled
areas (see Chapter 4), and detection probabilities are low. At the point of writing, six
indigenous communities living within the boundaries of the protected area have
received official communal land titles. With the exception of the initial pilot village,
assistance for land use planning and land titling has been prioritised in villages near
the district capital because land and forest resources were judged to be under the
greatest threat, as they are located close to good quality roads and, therefore, subject
to both high immigration and increasing commercialisation of agriculture. This has
added to the institutional complexity across the study site, as different villages are
now at various points along the tenure reform process.
The area is characterised into three livelihood zones in which different livelihood
activities dominate: a cash crop zone, a lowland paddy zone and an upland zone.
These zones have repeatedly been referred to in project documentation and reflect the
major inter-community heterogeneity with respect to bio-physical characteristics,
institutional framing, opportunity costs of stopping deforestation and economic wellbeing. As such, the majority of variation in clearance behaviour due to the conditions
in which decisions are made is expected to be represented by these zones. The cash
crop zone is centred around the district capital of Keo Seima and is characterised by
easy road access and mature cash crop markets. This area is also currently
experiencing the highest rate of land conversion (WCS, 2013). The lowland paddy
zone is located in the most remote part of the protected area. Access to this area is
difficult (particularly during the wet season) and the dominant livelihood strategy is
centred on the cultivation of paddy rice, supported by liquid resin collection from
native dipterocarp trees. The upland zone is also located further from market centres,
although access is largely better than for the paddy zone. In this zone, households
cultivate a greater diversity of crops and have recently made the transition towards
commercial production.
Variation in economic well-being between the three livelihood zones has previously
been assessed using a participatory poverty index, which was calculated following the
basic necessity survey method (see Chapter 3 for an explanation of this method).
Across the 622 households sampled, the household poverty score ranged from 0.10 to
109
0.82, with a mean of 0.44 (SE = 0.13). After correcting for other household
demographic and socio-economic variables, a significant difference was found
between the mean poverty score of households from villages in each of the three
livelihood zones. As expected, communities in the cash crop zone had the highest
mean household poverty score (0.10 higher than the mean score for villages in the
paddy zone (p < 0.001) and 0.04 higher than the mean score for villages in the upland
zone (p = 0.054)), indicating that these households are on average better off than
households from villages in the other two zones. The livelihood zone with the lowest
mean household poverty score was the lowland paddy zone (mean score for villages
was 0.06 lower than the mean score for villages in the upland zone (p = 0.016)),
which again was as expected as this is the most remote zone with fewer income
generating livelihood opportunities available to the people who live there. Within
each zone, there was little difference in heterogeneity at the household level. The
standard errors for household poverty scores in all three zones are comparable, with
the least variation in the upland zone and the most in the cash crop zone (SEcash = 0.14,
SEpaddy = 0.14, SEupland = 0.12).
Methods
All households surveyed as part of this research had previously participated in the
basic necessity survey. As such, it was possible to stratify each household by two
variables: poverty score and livelihood zone. Sampling for the scenario interviews
was spread equally across poverty score terciles, which were derived for the whole
survey area, and the three livelihood zones, such that an equal proportion of
households from the three poverty score terciles were interviewed for each zone. In
total, 49 households were interviewed: four households to pilot the method and
another five households for each poverty score tercile in each livelihood zone. In
cases where the household head was not present in the village at the time that the
survey was carried out, a replacement household was selected randomly from a list of
all households within the village. Rather than sample households from each of the 20
REDD+ villages, two or three villages were selected for each livelihood zone in such
a way as to cover the full range of conditions found in villages within each zone.
Interviews were semi-structured, with respondents presented with a baseline business
as usual scenario and a further six scenarios in which one aspect of future conditions
110
was changed. Each scenario was presented one by one and the respondent asked to
qualitatively explain how they would react over the next five years with regards to
their agricultural and other livelihood activities. Follow-up questions were asked to
ensure that the respondent had fully understood the scenario, to explore responses in
greater detail and to triangulate between responses to different scenarios. As land-use
is a potentially sensitive issue in Cambodian protected areas, respondents were
assured that all data collected would be confidential. A subsample of responses from
two villages was validated by comparing landholdings reported during the scenario
interviews with those reported in Chapter 4, with all responses comparing favourably.
Interview responses to the six future scenarios were compared against those to the
baseline business as usual scenario. Responses were then coded into different
categories depending on their likely impact on forest cover. In certain cases, in which
direct changes in clearance behaviour were not described, coding was based on
existing understanding and knowledge of land use practices within the study area. For
example, where respondents intended to buy land already in use, this was coded as a
resultant increase in forest clearance based on existing understanding of clearance
processes in sampled villages. In this case, it is reportedly common practice for
individuals to clear forest specifically for sale or to sell land within village boundaries
but subsequently clear additional forest to compensate for the loss (Milne, 2013). In
both instances, the act of buying land fuels additional indirect clearance. Where the
conditions presented within a scenario were contingent on the behaviour of others,
interview respondents were asked to describe their response to the scenario based on
whether others in their group cleared. Responses were then coded as conditional or
unconditional depending on whether they were contingent on the behaviour of others
or not.
There follows a description of the conditions presented by each scenario. A priori
hypotheses are included for each of the six scenarios in which the conditions were
different to those of the business as usual baseline scenario and are based on prior
understanding of behaviour within the study villages and existing literature.
Baseline business as usual scenario
In this scenario, it was explained that current conditions would stay constant over the
next five years. All average prices would remain at a stable level, although there
111
would be some small variation between each year. Law enforcement effort would
remain constant but, in cases where indigenous land title had yet to be granted, tenure
reforms would progress as currently planned.
Performance payments
In these three scenarios, respondents were offered the option of receiving an annual
payment of $200 in return for agreeing not to clear any additional land, including
inside areas designated for use under communal land title. Payments were set at $200
as this was felt to be representative of the likely sum offered in the future as part of
the REDD+ project (T. Evans, pers. comm.). In the first of these scenarios, payments
were to be made on an individual basis and were dependent solely on the compliance
of the household interviewed. In the second scenario, payments would be made based
on the behaviour of groups of ten individual households from within the same
neighbourhood as the respondent household. Each household would receive $200 if
all households in the group chose not to clear any additional land, otherwise they
would receive nothing. In the third of these scenarios, the money would be paid into a
village level development fund to support infrastructural development or other
development projects identified by the community instead of the $200 being paid to
each individual household. These payments would again be dependent on the
collective behaviour of groups of ten households from the same neighbourhood.
Hypothesis H6.1: the offer of conditional payments will reduce the number of
households who predict that they will expand their agricultural land relative to a) the
business as usual baseline, and b) the increased enforcement effort scenario.
Hypothesis H6.2: the offer of conditional payments will more strongly reduce the
number of households who predict that they will expand their agricultural land when
payments are conditional on group-level compliance than when they are conditional
solely on individual compliance, but the type of collective payment offered will not
make a difference to land expansion.
Increase in enforcement effort
In this scenario, respondents were told that enforcement effort within the study site
would be doubled, such that it would be twice as likely that illegal land clearance
would be detected. The likelihood of prosecution once caught was unaffected. This
112
additional enforcement activity would be restricted to conservation areas, i.e.
clearance inside titled areas designated for communal agricultural use would not be
subject to increased law enforcement as these areas are subject to community
management.
Hypothesis H6.3: doubling the probability of detection will decrease the number of
households who predict that they will expand their agricultural land relative to the
business as usual baseline.
Increase and decrease in cassava price
For these two scenarios, respondents were asked to imagine that the price of cassava
halved or doubled over the next five years, with some small yearly variation. Cassava
has quickly become the dominant commercial crop grown in most areas of the study
site and many farmers report that the price at which they are able to sell this crop is of
particular importance in determining future land use (H. Travers, pers. obs.). However,
the cassava market in the project area is imperfect and prices can be volatile. These
two scenarios were included to explore the effect that a consistent increase or
decrease in average cassava prices would have on future smallholder land use and to
compare the effect of such changes against those resulting from conservation
interventions.
Hypothesis H6.4: doubling the price of cassava will increase the number of
households who predict that they will expand their agricultural land relative to the
business as usual baseline.
Hypothesis H6.5: halving the price of cassava will decrease the number of households
who predict that they will expand their agricultural land relative to the business as
usual baseline.
Statistical analysis
A series of generalised linear mixed effects models were constructed to analyse the
probability of pro-conservation behaviour under the different scenarios considered.
Two response variables were considered. The first of these was a dichotomous
variable describing whether or not a household would follow pro-conservation
behaviour under the conditions presented by each scenario. In this context, pro-
113
conservation behaviour was defined as reducing clearance relative to the baseline
business as usual scenario, but including those who did not clear forest in either the
treatment or baseline scenario. The second response variable, a subset of the first, was
a dichotomous variable describing whether or not a household stated it would reduce
its clearance relative to the baseline scenario (i.e. excluding those who did not clear in
the baseline scenario). For both response variables, models were considered for
conditional (those dependent on reciprocation) and unconditional (those independent
of reciprocation) responses. Model selection was carried out using backwards
stepwise selection on the basis of corrected Akaike information criterion (AICc;
Akaike, 1979) values. Following Burnham and Andersen (2002), the most
parsimonious candidate models with a ΔAICc value of less than two were selected,
otherwise the model with the lowest AICc was selected. In each model, an individual
random effect was included to account for multiple responses by each individual.
Inclusion of a village random effect was tested using likelihood ratio tests and
rejected for all models (Bolker et al., 2009). Explanatory variables included livelihood
zone, poverty score and demographic variables, such as age, sex and ethnicity (see
Appendix D for details).
Results
Under the baseline business as usual scenario, 82% of respondents declared that they
would continue to increase their agricultural land over the next five years, while the
remaining 18% stated that they had no intention of expanding their land. This shows
the high level of smallholder farmer involvement in forest clearance in the study site.
The stated reactions of each household to the scenarios presented were coded relative
to those under the baseline scenario (Figure 6.1). This showed significant variation in
the reported responses to each scenario, ranging from a large increase in the rate of
forest clearance under increased cassava prices to a potentially large decrease under
collective payments or a village development fund. Overall, however, a high
proportion of respondents reported that their behaviour would be unaffected by the
conditions presented to them in a given scenario. This proportion was particularly
high for both the enforcement and individual payment scenarios, for which only 11%
of respondents reported that they would change their proposed clearance behaviour
relative to business as usual. The results of the logistic mixed effects models of pro-
114
conservation behaviour and reduced clearance under different scenarios confirm many
of the observational results above (Table 6.1). The scenario presented was a
significant determinant of stated behaviour in both models.
Figure 6.1 Proportion of coded responses for the business as usual baseline scenario and each of
the six future scenarios. Conditional behaviour refers to behaviour that is contingent on
reciprocation.
Options for benefit sharing
With respect to the different options considered for benefit sharing, the two collective
payment scenarios performed significantly better than individually contracted
payments, provided that others in the group reciprocated. The probability of behaving
in a pro-conservation manner increased relative to the baseline scenario by 0.53 for
the collective payments scenario and 0.35 for the village fund scenario, but only by
0.03 for the individual payment scenario (Table 6.1, column 1). When only
unconditional responses were considered, all of the benefit sharing scenarios
considered had only a negligible effect on response relative to the business as usual
baseline, even though the clearance rate reduced significantly under the village fund
scenario (Table 6.1, column 2). With respect to reductions in forest clearance,
collective payments significantly out-performed individual payments. With payment
into a village fund as the reference, the probability that a household would reduce
clearance under individual payments was 0.58 lower than for collective payments
(Table 6.1, column 3). The probability that individual payments would reduce
household agricultural expansion was only 0.05. As such, there is mixed support for
hypothesis H6.1a but strong support for hypothesis H6.2.
115
Table 6.1: Parameter estimates for the pro-conservation and clearance reduction models considering conditional and unconditional responses. The difference in the
probability of acting in a pro-conservation manner or reducing clearance relative to the baseline scenario was found by calculating the expected probability for
each variable and comparing this with the probability expected for the intercept. The full list of variables modelled, plus their explanations, are given in Appendix
D; Table D.1. Significance values: ns = not significant; (*) = P < 0.1; * = P < 0.05; ** = P < 0.01; *** = P < 0.001. Reference level treatment indicated by ref.
Variable
intercept
scenario:
baseline
cassava up
cassava down
enforcement
individual
collective
fund
zone:
cash
upland
paddy
random effect:
individual (SD)
AICc
N
Pro-Conservation Model
conditional
unconditional
Clearance Reduction Model
conditional
unconditional
0.007 (0.010) ***
0.001 (0.003) ***
0.435 (0.121) ns
0.041 (0.040) ***
ref
-0.006 (0.002) *
0.067 (0.072) **
0.026 (0.035) *
0.026 (0.035) *
0.532 (0.186) ***
0.353 (0.192) ***
ref
-0.001 (0.000) *
0.021 (0.029) ***
0.006 (0.010) *
0.006 (0.010) *
0.004 (0.007) *
0.016 (0.023) ***
-0.430 (0.011) ***
-0.327 (0.071) ***
-0.389 (0.040) ***
-0.389 (0.040) ***
+0.191 (0.114) ns
ref
-0.040 (0.004) **
0.011 (0.050) ns
-0.023 (0.020) ns
-0.023 (0.020) ns
-0.023 (0.020) ns
ref
0.535 (0.180) ***
0.019 (0.025) (*)
ref
0.584 (0.252) ***
0.008 (0.022) (*)
ref
1.387
253.8
315
2.213
210.8
315
2.054
234.1
270
2.906
182.0
270
116
As with pro-conservation behaviour, the probability that households would reduce
clearance relative to the baseline scenario was significantly higher for both collective
reward scenarios if conditional behaviour was considered than when only
unconditional behaviour was considered (Table 6.1, columns 3 and 4). As such, these
results show that the offer of collective payments to individual households or to a
village development fund had the potential to be the most effective benefit sharing
option with respect to reducing forest clearance but, crucially, a high proportion of
respondents stated that their decision to stop or reduce clearance was conditional on
the cooperation of others. Within this group the extent of their cooperation was
roughly evenly split into two categories: those households who would only stop
clearing provided that most or all others in their group reciprocated and those who
would only continue to clear if most or all others in their group did likewise.
Consequently, whilst both scenarios offer a potentially significant reduction in
clearance, there is a risk that compliance with the conditions of the payments would
be undermined: first by the minority of respondents who rejected the payment outright
and subsequently by those respondents who would only cooperate if everyone else in
the group did so.
Responses to the individual payments treatment were similarly nuanced. While the
majority of respondents rejected the offer of $200 in return for not clearing, the
reasons for doing so varied. For many, the equation was simple: the $200 payment
offered was insufficient to compensate them for the opportunity costs they would
experience by halting agricultural expansion. Another commonly voiced reason was
that if respondents did not clear any more land, the land currently in use would
gradually lose fertility and at present they had insufficient land to rotate cultivation. In
these cases, some respondents reported that they would be happy to agree to the
conditions of the payment once they had secured sufficient land. In other cases,
however, respondents rejected the conditions of the payment even though the
projected earnings of the land they proposed to clear (based on current earnings from
the land they held already) were lower than the payment offered. For these households,
the reasons for clearing related less to the potential income that they could generate
from the land but rather to securing land for their family for the future. Concerns were
frequently raised that unless they cleared the five hectares allowed under community
rules as part of the communal land titling process, the land was more vulnerable to
117
seizure from outside interests, a significant issue across the country. A similar concern
related to whether there would be sufficient land available for their children when
they married (i.e. whether the allowance made under communal titling for future
expansion of the village was sufficient) and that it was better to clear the land now to
secure their children’s future. In these cases, the opportunity cost of not clearing
related less to the potential annual income that could be generated from the land than
to perceived failures in the long-term management of the land titling process.
Enforcement vs. benefit sharing
One of the decisions facing PA authorities in Seima is how best to invest REDD+
revenues to reduce forest clearance. To this end, investment in the most effective
benefit sharing option, collective payments, was compared with additional
enforcement effort. The results of the pro-conservation model show that doubling the
detection rate for people who clear forest illegally resulted in a negligible effect on
response (0.03 increase in the probability of acting in a pro-conservation manner
relative to the baseline; Table 6.1, column 1). The probability that a household would
reduce agricultural expansion as result of increased enforcement effort was estimated
to be 0.04. This provides only very weak support for hypothesis H6.3 that doubling
the rate of detection would reduce the number of households expanding their
agricultural lands relative to the business as usual baseline. In this regard, the
enforcement scenario performed poorly in comparison to collective payments
(probability of reducing clearance 0.39 lower than for collective payments; Table 6.1,
column 3). As such, assuming that cooperation is reciprocated amongst groups under
the collective payment scenario, increasing enforcement effort was found to be a poor
alternative to benefit sharing, which provides strong support for hypothesis H6.1b.
One of the reasons behind the lack of effectiveness of the enforcement scenario lies in
the recent land tenure reforms that have been implemented throughout the study area.
In all but one of the villages sampled, indigenous communal land title has been
granted or is in the process of being applied for. Consequently, land has been
identified for future expansion of agricultural activities in each village. As such, any
infraction within this land is not subject to law enforcement by protected area staff,
but governed by community land use rules drawn up by the overseeing village
committee. Interview respondents were, therefore, able to claim that they would be
118
unaffected by increased law enforcement effort as none of their proposed expansion
would encroach conservation areas.
Conservation interventions vs. exogenous price changes
Another issue facing managers at the study site is the rapid commercialisation of
agriculture. While this has resulted in significantly increased household incomes in
the areas that have benefited from it (Chapter 3), the rate of change threatens to
swamp interventions aimed at controlling the expansion of agricultural land.
Respondents were presented with only one change from business as usual in each
scenario in order to keep the exercise simple for the respondents (i.e. no scenario
contained both price changes and a new conservation intervention). Despite this, it is
informative to compare responses under the exogenous price changes scenarios and
the most effective conservation intervention, collective payments.
Whilst the two scenarios that presented exogenous changes related to the sale price of
cassava resulted in negligible differences to the probability of acting in a proconservation manner relative to the business as usual baseline (-0.01 for the cassava
up scenario and 0.07 for the cassava down scenario; Table 6.1, column 1), the picture
this paints is slightly misleading. For the scenario in which the sale price doubled,
over 40% of respondents reported that they would increase their intended clearance
relative to the business as usual baseline, which is comparable to the 58% of
respondents that said they would reduce clearance under the collective payments
scenario assuming others reciprocated. A chi-squared test of responses to the cassava
up scenario shows that a significant proportion of the population would increase
clearance relative to business as usual if the price of cassava doubled (chi-squared test,
df = 1, p = 1.5e-6), providing strong support for hypothesis H6.4. Not only does this
illustrate the significant impact that this one crop has on the landscape, but it also
highlights the risk of the positive effects of investment in benefit sharing being
undermined by exogenous changes in the study villages. There are also implications
for any potential livelihood intervention aimed at improving the profitability of the
cassava crop as a way of reducing the need to deforest, as there is a distinct risk that
this would perversely increase the incentive to clear land. It is also worth noting that
the scenario in which the cassava price halved had the highest proportion of all
scenarios of respondents unconditionally stopping or reducing their proposed
119
agricultural expansion, which supports hypothesis H6.5. As such, an exogenous price
reduction resulted in greater reductions in clearance than both the enforcement and
individual payment scenarios. Conditional responses to the collective payments still
led to significantly higher reduction in clearance than that resulting in price reductions,
however. The majority of respondents reported that their income would suffer under
cassava price reductions, as alternative livelihood options available to them are less
profitable.
Heterogeneity of Response
Understanding how the response to different scenarios varies with changing
conditions and household characteristics may help to identify opportunities for
targeting interventions in specific areas or at different groups. Separating scenario
interview responses by livelihood zone revealed significant differences between the
responses of households in each zone, with households in the cash crop zone reporting
very different behaviour to the other two zones (Figure 6.2, row 1). The greatest
difference resulted from over 50% of the respondents in this zone reporting that they
would not clear or buy any further land under business as usual conditions, either
because they felt that they already had sufficient land, there was no land available to
be cleared or they did not want to clear land illegally.
As expected from the observational results, the logistic mixed effects regression found
pro-conservation behaviour to be significantly affected by livelihood zone, with the
probability that households from the cash crop zone would act in a pro-conservation
manner predicted to be 0.54 higher than households in the remote paddy zone (Table
6.1, column 1). Unlike the conservation interventions considered, this effect was
maintained when only unconditional behaviour was considered (0.58 increase in
probability; Table 6.1, column 2). This result comes from households stating that they
had no plans to expand their fields even under the baseline scenario and so would
behave in a pro-conservation manner no matter which intervention was put into place.
From the perspective of understanding which interventions would be the most
effective at reducing clearance and where those interventions should be targeted to
have the greatest impact, this is not particularly helpful. However, livelihood zone
was not included as an explanatory variable in the selected model for the two models
that considered whether households would actively reduce clearance relative to the
120
baseline scenario. This suggests that, even though pro-conservation behaviour is
predicted to be higher in the cash crop zone, there is no zonal effect on the
effectiveness of different interventions with respect to active reductions in forest
clearance.
In contrast to separating responses by livelihood zone, very little difference was
apparent between the response of households from different basic necessity survey
terciles (Figure 6.2, row 2). Household poverty score and other demographic variables
also did not feature in the selected models explaining clearance behaviour. This
suggests that differences between the conditions in the three livelihood zones may be
a more important factor influencing decision-making than individual household
heterogeneity.
121
Figure 6.2: Proportion of coded responses for the business as usual baseline scenario and each of the six future scenarios stratified by livelihood zone (row 1) and
household poverty score terciles (row 2). Conditional behaviour refers to responses that are contingent on reciprocation.
122
Discussion
Intervention effectiveness
The results of the scenario interviews show that investment of REDD+ revenues can
dramatically reduce the number of smallholder households involved in agricultural
expansion, and in turn reduce clearance of biodiversity rich forest. Both of the
collective benefit sharing options (collective payments and payments into a village
fund) resulted in significant reported reductions in agricultural expansion. The caveat
to this result, however, is that these reductions were only observed if conditional
behaviour was included in the analysis. This finding is similar to several previous
studies of collective action which have found that within a population, whilst a
minority will either chose to cooperate or not in most situations, the majority are
classified as ‘conditional cooperators’, who will only cooperate provided that other
people reciprocate (Fischbacher et al., 2001; Rustagi et al., 2010). This suggests that
compliance with the terms of a conditional payment would be dependent on
facilitating a strong sense of collective action within each village, possibly by
maximising opportunities for dialogue within the community by undertaking a
genuinely participatory approach to designing the benefit sharing system.
The results also demonstrate the risk of investing in activities that may prove
ineffective at changing the incentives for households to reduce expansion. Neither
doubling current levels of law enforcement, nor offering payments to individual
households had any meaningful effect on pro-conservation behaviour and both
performed poorly in comparison to the two collective benefit sharing scenarios at
reducing overall clearance. With respect to enforcement, the results of the scenario
interviews confirm previous studies, which suggest that the effectiveness of law
enforcement in Cambodia is weak in situations where expected penalties are low
(Claridge et al., 2005). This is unsurprising in a country such as Cambodia, where the
“rule of law” generally is weak and corruption levels high (Kheang, 2006). It is
possible, however, that the effectiveness of increasing law enforcement effort may
have been underestimated, as respondents were able to claim that they planned all
future clearance inside community boundaries. As already shown by the results
presented in Chapter 4, compliance with community boundaries is far from perfect
even amongst households with less than 5 ha (i.e. those households with less than
123
their allocation for land within community titled areas). Hence, respondents may have
been over-reporting their intention to clear in titled areas. With respect to individual
incentive payments, these have been found to be more effective than collective
payments in some other contexts (e.g. Narloch et al., 2012). From this previous work
it might easily be assumed that individual payments are a viable or superior
alternative to collective payments for Seima, yet this has been shown not to be the
case in this instance. This result highlights the value of using approaches, such as
scenario interviews, to investigate possible responses to newly introduced policies
prior to their implementation in different contexts.
Finally, the results of the two scenarios in which the price of cassava was changed
offer some perspective on the risk of assuming the continuation of business as usual
conditions when planning conservation interventions. Until 2005, cassava made up a
negligible proportion of total cultivated area in the study site. Since that time, the area
of land under cassava cultivation has increased exponentially and many households
now only grow cassava (see Chapter 4). The unforeseen introduction of a new cash
crop has, in just a short space of time, had a dramatic influence on both the landscape
and people’s livelihoods. This is a salutary lesson in how quickly the context in which
conservation interventions are implemented can change. In this research, the choice of
scenario was restricted by possible futures that interview respondents could conceive
of sufficiently in order to be able to predict their behaviour, which precluded
unfamiliar changes. However, from the observed effects of increasing the price of
cassava it is evident that exogenous changes may bring about differences in future
clearance rates that are comparable to those achieved by planned interventions.
Similarly, planned changes to the profitability of cassava, for example through
proposed agricultural intensification (Milne et al., 2012), could result in perverse
effects as has been found in other contexts (Angelsen & Kaimowitz, 1999; Damania
et al., 2005)
Intervention interaction with communal land titling
One of the challenges facing the developers of the REDD+ demonstration project in
Seima is the design of benefit sharing with respect to existing activities, such as the
conservation-led indigenous communal land titling process. In the majority of villages
within the study area, community managed agricultural areas have either been
124
approved or applied for. These areas are designated for agricultural purposes and yet,
in many villages, the main land cover within these areas is still forest. Clearance
within these areas will, therefore, count against the REDD+ project and reduce
revenues. In the benefit sharing options considered in this research, benefits were
made conditional on households ceasing all clearance, irrespective of where that
clearance might take place. This has the advantage of disincentivising all clearance
but carries with it the risk that some households will feel it necessary to expand their
land anyway because they have insufficient land to provide for themselves. An
alternative option, in which payments are only contingent on not clearing outside of
the indigenous communal land title, would allow households to clear up to 5 ha within
titled areas, essentially writing off the forest within these areas, but would provide
clear incentives for title boundaries to be respected3. This option would also improve
integration with law enforcement activities, which focus only on clearance outside of
the community areas.
This ties in strongly with the differences in responses for households from different
livelihood zones. In the cash crop zone, tenure reform has progressed the furthest,
with the majority of villages having now received communal title. Villages in this
zone have also experienced the greatest level of encroachment on their land and, as
such, there is now little legal land available in these villages. Much of the clearance
and agricultural expansion in this area is now, therefore, illegal. These conditions are
likely to reduce the available options for households that are unwilling to break the
law and may not be able to afford the high land prices that result from the scarcity of
available land. It is, however, doubtful that clearance rates in this zone would be
lower in reality, as much of the clearance experienced is due to migrants moving to
the study area. For the remaining two zones, the majority of villages have only
recently had boundaries agreed as part of their application for communal tenure,
particularly in the case of the paddy zone. Commonly expressed sentiments were the
need to clear to protect community land from outside interests and the right of each
household to cultivate up to five hectares of land. This is reflected both in the low
3
Subsequent to the completion of this chapter, a Verified Carbon Standard audit of
the REDD+ project required that areas under communal management, including all
ICT areas, be withdrawn from project boundaries, as suggested here.
125
number of responses that indicated an unconditional reduction in clearance in
response to the different scenarios and by the 100% of respondents in these two zones
reported their intention to expand their land holdings under the baseline scenario.
Fears regarding their security of tenure are driving households to clear more land than
they might otherwise. As such, it may be better to make benefits conditional only on
clearance outside community areas, which would allow households to secure their
titled land from possible land grabbing in the future. If payments are made conditional
on halting all clearance, they may be perceived as being unfair, even if they are
voluntary.
Despite some obvious differences in responses between the three livelihood zones, no
significant difference was found with respect to the effectiveness of different
interventions to reduce agricultural expansion. Similarly, no effect on response was
found for different household characteristics, including household poverty score. This
consistent response across different zones and households suggests that, in this
instance, few opportunities exist for targeting interventions in specific places or at
specific groups. Whilst this simplifies implementation, as the same level of benefits
can be offered to every household across the landscape, these results suggest that
there will be some households that, initially at least, will choose not to cooperate with
the REDD+ project and will continue to clear no matter what conservation
intervention is implemented. The key, however, will be to mobilise conditional
cooperators to support the conservation intervention.
126
Chapter 7
7. Discussion
Benefit sharing for REDD+
“Power tends to corrupt, and absolute power corrupts absolutely”
Baron Acton (1834–1902)
One of the principal narrative strands relating to the implementation of REDD+
regards the opportunities for the delivery of co-benefits for both biodiversity and local
livelihoods alongside the primary goal of reducing carbon emissions (Brown et al.,
2008; Stickler et al., 2009), such that successful projects could result in win-win-wins.
This is not an easy task and the history of conservation is littered with poorly
conceived policy initiatives that have failed to balance the need for trade-offs between
seemingly conflicting objectives (Blom et al., 2010, McShane et al., 2011). In this
study, I have focussed on analysing the possible future effects of different
interventions under consideration at one REDD+ site in order to understand which of
these would have the greatest impact on resource use if implemented. This process
has generated results that will directly feed into the design of conservation activities at
that site, yet there are wider lessons to be learnt.
The most widely applicable lesson regards the structure of performance payments as
part of a benefit sharing system, and the institutions that are created or adapted to
manage them at the local level. The results of Chapters 5 and 6 both suggest that
collective payments are preferable to individual payments either outright (Chapter 6)
or because of associated added benefits, such as carry over effects or positive
interactions with enforcement (Chapter 5). Responses to the scenario interviews in
Chapter 6 also suggested that collective payments were most effective when disbursed
as cash rather than contributions to a village-level development fund. These predictive
results are supported by the findings of recent work undertaken in protected areas in
the Northern Plains Landscape of Cambodia, which investigated the effectiveness of
pilot incentive schemes similar in structure to those considered here (Clements et al.,
2010; Clements, 2012; Clements & Milner-Gulland, in press).
127
These are useful results that can help guide policy setting for REDD+ projects, yet
there are also implications for how such payments are managed. One of the reasons
payments were preferred to contributions was a widely held scepticism that a village
fund, managed by the indigenous community committee in each village, would be
free from corruption or invested in projects that would benefit everyone. This echoes
the results of Chapter 4, which showed that members of the management committee
in O Rona were amongst the worst offenders with respect to non-compliance with
community land use regulations and were regarded poorly by other members of the
indigenous community. This issue cuts deeper than simple elite capture of resources, a
phenomenon commonly encountered by community style interventions (Mansuri &
Rao, 2004; Platteau, 2004). In Seima, the Bunong have strong traditional institutions,
which are tied to their animist belief system and interactions with the natural
environment (UNDP, 2010). It is convenient for conservation interventions that
require management at the local level, such as land use planning or village
development funds, to link into these institutions, as it is commonly perceived to
afford them added legitimacy (Agrawal & Gibson, 1999; Berkes, 2004). Yet, this
approach fundamentally alters the role of traditional village authorities, which can
dramatically change the way in which they are perceived (as seen in O Rona; Chapter
4). Consequently, there is a risk that the implementation of conservation interventions
can act to undermine traditional authority and erode cultural well-being (Jones et al.,
2008); impacts that may go unnoticed, as they are particularly difficult to quantify or
measure (Milner-Gulland et al., 2014). Conversely, collective benefits have, in some
instances, been found to strengthen local institutions (e.g. Clements et al., 2010).
Another lesson regards the extent to which interventions should be targeted at specific
groups. Part of the value of approaches such as economic games (Chapter 5) and
scenario interviews (Chapter 6) lies not only in being able to assess the relative
effectiveness of different policy options but also in identifying sections of a
population that may be particularly responsive to interventions. However, beyond the
headline results of which intervention might bring about the greatest behavioural
change, the results of Chapters 5 and 6 showed that there was a great deal of
consistency observed in responses between groups for each approach. While there is
significant diversity across the study site with respect to livelihood opportunities,
economic well-being, ethnicity and culture, the two approaches found little evidence
128
to suggest that response to the different benefit sharing options considered would vary
across the population. This was found to be true even in situations where overall
behaviour differed, such as the difference in pro-conservation behaviour observed
between different livelihood zones in Chapter 6. This is encouraging because it
suggests that, in this instance, an egalitarian approach to benefit sharing, in which the
same benefits are offered to every household, is appropriate. This is important, as it
runs counter to the commonly heard suggestion that approaches that improve equity
outcomes, which can add legitimacy to conservation efforts (Skutsch et al., 2011),
sacrifice efficiency (Alix-Garcia et al., 2008; Proctor et al., 2009; Pascual et al., 2010).
However, the results of Chapter 3 show that, while overall household economic wellbeing increased from 2007 to 2012, certain groups remain disadvantaged.
Consequently, there is an argument for focussing benefit sharing on vulnerable groups,
or at the very least ensuring that such groups do not lose out.
Finally, since the increased interest in conserving forest resources generated by the
emergence of REDD+, there have been fears of a revival of green-grabbing, and a
recentralisation of forest management (Phelps et al., 2010). These fears have been
supported by the emergence of evidence that the rate of decentralisation of forest
management has slowed significantly in recent years (RRI, 2014). Yet, in Seima,
where indigenous communities are receiving title over their customary lands (as
described in Chapter 4), titled areas are being removed from the REDD+ project
because the project applicant (the Cambodian government) does not have full
management rights to that land. This has created a bizarre situation in which project
proponents have followed widely held recommendations to secure legal recognition of
the customary rights of local people and obtained free, prior, informed consent from
all villages identified as having land or extracting resources from within the project
boundary, yet those areas have now been excluded from the project. As such, any
incentives created to reduce clearance by local people must focus on illegal expansion
beyond titled areas. It may be that this would be a more effective approach than
providing incentives to reduce clearance inside titled areas (as discussed in Chapter 6),
but it reopens the debate about the equity of benefit sharing because households living
in participating villages, in effect, have no more right to exploit resources within
project boundaries or receive benefits from the project than immigrant households
moving to the area from other provinces.
129
Experimental games vs. scenario interviews
Two of the approaches used in this study, experimental games (Chapter 5) and
scenario interviews (Chapter 6), evaluated the potential effectiveness of interventions
under consideration at the study site. While the application of neither of these
approaches to investigating the likely effects of conservation interventions is entirely
novel, no other study has applied both approaches to the investigation of policy
effectiveness in the same contextual setting. It is instructive, therefore, to look at how
each of these two approaches performs in comparison with the other, and consider
whether the combined results allow for greater confidence regarding the external
validity of the individual approaches and conclusions.
Combined, the two approaches offer considerable complementarity. One advantage of
the qualitative approach taken during the scenario interviews is the ability to follow
lines of enquiry as they develop during the interviews. The experimental setting of the
CPR game allows for fewer opportunities to query the thought behind individual
decisions. Although each participant was interviewed at the end of every experimental
session, they often found it difficult to remember how they felt during each round or
explain their motivations behind each decision. The scenario interviews, therefore,
offer greater scope for interrogating the reasons why responses to the different
interventions vary. Scenario interviews also allow for greater complexity to be
included in the policy context considered, as descriptive narratives are often easier to
understand than the stylised conditions of experimental games. For this study, this
meant that land clearance, which was considered too complex to be understandable in
the context of the CPR game, could be investigated using scenario interviews. On the
other hand, the results of the CPR game allow for direct comparisons in the
magnitude of response to different policy options, rather than simply the number of
households that would alter their behaviour. Combined, therefore, the two approaches
provide a framework to investigate future policy effectiveness in both a quantitative
and qualitative manner.
However, both of these two approaches have had criticisms levelled at them regarding
the external validity of the results obtained through their application. While
experimental games have been increasing applied to understanding behaviour within a
commons or other social dilemma, there have been surprisingly few studies that have
130
investigated external validity in an empirically rigorous manner (Levitt & List, 2007;
Martin, 2013). Following the taxonomy of experiments devised by Harrison and List
(2004), framed experiments are considered to have greater validity than artefactual
experiments and field experiments greater validity than lab experiments. Framed field
experiments, such as described in Chapter 5, are therefore expected to offer greater
validity than experiments involving volunteers in a lab setting. Yet, the limited
evidence that exists in regard to experimental external validity is mixed. For example,
in one of the most controlled comparisons between real life and experimental
behaviour made to date, Voors et al. (2012) found no meaningful correlation between
behaviour in a context-free public goods game and response to a real life development
intervention in the same subject pool of individuals in Sierra Leone. Yet, in a
comparison between behaviour in a framed CPR game and three real life incentive
payment schemes, Martin (2013) found that those individuals who were characterised
as co-operators or conditional co-operators in the CPR game in Cambodia were less
likely to clear land illegally (and, thus, break the terms of the PES agreements) than
those individuals characterised as free-riders from their behaviour during the game.
As such, doubts remain as to whether or not the results of experimental games should
be used to guide policy decision-making.
For scenario interviews, the major concern relates to the hypothetical nature of the
approach and the potential for disparity between stated and observed behaviour.
Participants in an experimental game are placed in a real social dilemma, no matter
how stylised, in which the decisions they make will affect the size of payoff they
receive at the end of the experiment; an approach that is considered critical to the
validity of the results (Croson, 2005). Yet, this control is lacking for scenario
interviews. The potential for disparity between observed behaviour and that stated in
an interview setting has been particularly investigated for contingent valuation (e.g.
Neill et al., 1994; Brown et al., 1996; Loomis et al., 1996; Foster et al., 1997), a
method used to estimate an individual’s willingness to pay or accept different futures
from their stated preferences, with stated willingness to pay found to be higher than
observed willingness to pay in the majority of studies (Venkatachalam, 2004). Again,
this potential for misleading results may cast doubt on the suitability of using
responses to hypothetical scenarios to help guide policy making.
131
Given these questions regarding the validity of the two approaches, it is worthwhile
comparing the results presented in Chapters 5 and 6. For the three common policy
options considered, the results of the two approaches are closely matched for
increased enforcement effort and collective payments and produce different results for
individual payments. Of the matched results, increased enforcement effort was found
to be ineffective for both approaches while collective payments were found to
significantly reduce resource use, even though the size of the payment offered should
have been insufficient to offset forgone income from extraction. It is particularly
encouraging that the results of both approaches identify collective payments as being
the most effective at changing behaviour from that predicted under business as usual
conditions. Triangulated in this way, the fact that the two sets of results compare so
strongly lends confidence to the validity of the conclusions drawn.
However, before this triangulation of results can be accepted, it is also important to
consider the differences in the findings for individual incentive payments, which were
found to be largely ineffective using scenario interviews but significantly reduced
resource appropriation in the CPR game. Specifically, how might the differences in
approach used explain the apparent disparity between strong agreement for collective
payments and disagreement for individual payments? For collective payments, the
principal mechanism by which cooperation is elicited is through peer-pressure and
fear of disapproval or hope of approval. In the CPR game, this effect was directly
realised through disclosure of individual decisions to the rest of the group. During the
scenario interviews, while respondents were not directly subject to peer-pressure, they
were made aware of the implications of their decisions during questioning. For
individual payments, while there was no direct peer-pressure to act in any particular
way, participants in the CPR game were able to discuss their options with the rest of
their group and still faced public disclosure of their decisions. But respondents of the
scenario interviews made their decisions individually with neither assistance nor
disclosure. In this respect, the game participants were able to draw upon greater
collective decision-making than respondents to the scenario interviews. As group
based decision-making has been shown previously to lead to more advantageous
results for individuals (Kocher & Sutter, 2004), this could explain why the results of
the two approaches match for collective payments but differ for individual payments.
However, further evidence would be required to support this explanation.
132
Overall, then, the results of the two approaches offer significant complementarity but,
while there is some commonality in findings, there was insufficient agreement to be
able to triangulate results as a proof of the external validity of these methods.
Complexity in conservation policy research
“In the land of the blind, the one-eyed man is King”
Desiderius Erasmus (1466 – 1536)
One of the major challenges facing conservation policy makers is predicting the likely
impact of new interventions implemented within complex institutional settings
(Hirsch et al., 2010; Game et al., in press). The institutional context in which resource
use decisions are taken has a significant impact on outcomes, yet it is rare for the
effect of interactions between interventions, which add greatly to contextual
complexity, to be investigated. In this study, I have shown that this is an omission that
should be corrected in the future.
As discussed above, the interaction between the creation of land use boundaries under
indigenous communal titling and the effects of performance payments made
conditional on clearance reductions (presented in Chapter 6) have implications that go
beyond the effect on resource use. Similarly, the results of Chapter 4 underlined the
important interaction between the indigenous land titling process and law enforcement
efforts. In both of the villages studied in Chapter 4, there was significant illegal
household land clearance outside of titled areas, although the extent and proportion of
households involved were both significantly greater in O Rona. In O Rona, the land
titling process was also being undermined by the high level of immigration and illegal
land grabbing within community titled areas. These results highlight a failure by
protected area authorities, and the wider Cambodian legal system, on the one hand to
protect the legal rights of Seima’s indigenous communities and on the other to
prosecute those individuals guilty of illegally clearing protected forest. The
effectiveness of indigenous land titling, and participatory land use planning
approaches more generally, in reducing uncontrolled land clearance stems from two
principal tenets: security of tenure and defined boundaries clearly delineating areas
where clearance is permitted and where it is not. The results of Chapter 3 show that,
although many households attribute natural resource security to the conservation
133
programme, over 50% of households do not feel secure about their resources. If
security of tenure cannot be assured because communities are not provided with
assistance in protecting their land and encroachment of protected forest is allowed to
go unchecked, it becomes significantly more difficult to predict whether participatory
land use policies will be successful.
The results of the experimental games presented in Chapter 5 provide further support
for the need to consider interactions between interventions carefully, by illustrating
the potentially nuanced effects such interactions can have on resource user behaviour.
While it has previously been shown that weak enforcement of resource restrictions
can undermine existing norms, such that extraction increases relative to that
experienced in the absence of enforcement (Cardenas et al., 2000; Vollan, 2008), this
undermining effect has not previously been observed to reduce the performance of
positive incentives such as payments. The fact that this crowding out effect was only
observed when payments were offered to individuals, but not groups, further
demonstrates the potential pitfalls of considering interventions in isolation.
Given the potential for interactions between conservation interventions to add
contextual complexity or have unexpected effects on behaviour, coupled with the
general confounding effect that institutional complexity can have on policy setting,
there is cause to consider the trade-off in research between the desire for increased
contextual relevance and the risk of parochialism. In part this represents a trade-off
between the desire of researchers to be able to draw wide-reaching lessons from their
work and the requirements of practitioners, who want results to be directly applicable
to the context in which they are working. Several of the results presented in this study
are potentially limited to the institutional context found in Seima Protection Forest, or
even specific villages. For instance, the results of Chapter 4 show that the difference
in the conditions found in the two villages had a significant impact on household
compliance, and hence the success of indigenous land titling in reducing illegal
clearance. Whilst it is still possible to tease out lessons that apply more widely, there
is a risk that in so doing those lessons may not apply to other contexts. For example,
one of the main results of this study is to show that collective payments are more
effective at reducing resource use than individual payments, but in another context
Narloch et al. (2012) found that individual payments were more effective. Given the
134
difficulties in applying lessons learned from behavioural research to other contextual
settings and the potential benefits of in-depth studies, which are better able to take
account of the multi-faceted nature of local context (Drury et al., 2011), I would argue
that the charge of parochialism seems a small price to pay for gaining a more precise
understanding of behaviour, albeit within a more restricted context.
Opportunities for future research
The results presented in this thesis have for the most part focussed on conservationdriven institutional change to reduce resource extraction, particularly that of
smallholder land clearance. And yet, the mechanisms by which the interventions
considered affect the decision-making processes that lead smallholders to clear remain
poorly understood. In Seima, the decision regarding whether or not to clear additional
land is influenced by a wide variety of factors, including but not limited to labour
availability (both within and outside the household), tenure concerns and the
perceived profitability of agricultural land use and alternative livelihood options, as
well as the institutional settings in which the decision is taken. It is doubtful, however,
that these decisions are made completely rationally, but that individuals instead rely
on simple heuristics to help guide them (Simon, 1972). Increased understanding of
how land use decisions are taken at the household level, and the factors that are
considered, would give greater confidence in the results presented here regarding the
relative effectiveness of different policy options, and would also allow for further
refinement of the options considered.
One of the leading approaches with respect to understanding underlying decisionmaking processes is agent-based modelling, a computational modelling technique
which allows researchers to study a heterogeneous set of evolving, autonomous
decision-making entities (agents) that interact with their environment and other
decision-makers (Janssen, 2005). In allowing agents to interact in this way, agentbased models provide a potentially powerful tool for examining behaviour and,
consequently, have been used in a wide array of disciplines in the social and
environmental sciences (Matthews et al., 2007). In particular, the development of
multi-agent systems, which combine agent-based and cellular models, has heralded
new opportunities to investigate the underlying processes of land use change within
complex socio-ecological systems (Wooldridge, 2002).
135
One of the key choices facing researchers using agent-based models is how to define
the interaction between agents and with the environment (Janssen, 2005). A
commonly applied approach is to follow the theory of bounded rationality (Simon,
1972), which sets out the idea that people often rely on simple heuristics rather than
optimisations as the basis of decision-making. Similarly, agents often make decisions
on the basis of imperfect information and preferences can change as new information
becomes available through interactions with other agents and the environment
(Heckbert et al., 2010). Defining rules for agent interactions and strategies can,
therefore, be particularly challenging in complex socio-ecological systems, in which
individual decision-making can produce significant externalities for other
stakeholders. It is for this reason that multi-agent systems are increasingly combined
with empirical methods to define agent behaviour within the particular system of
interest. Robinson et al. (2007) list five methods of empirical methods that have been
used in conjunction with agent-based models; sample surveys (Berger &
Schreinemachers, 2006), participant observation (Huigen, 2004), field and laboratory
experiments (Castillo & Saysel, 2005; Evans et al., 2006), companion modelling
(Castella et al., 2005; Barnaud et al., 2008) and GIS and remotely sensed data
analysis (Deadman et al., 2004). These methods differ significantly in their approach,
scale and ability to supply data on social and biophysical processes but can all be used
to provide information relating to macro-level phenomena and micro-level processes.
At the macro-level, this information is most widely used for comparison with model
results as a means of validation, whereas micro-level information is collected with the
intention of informing model coding.
The empirical approaches used in this study - household surveys (Chapter 3), GIS
analysis of land use change (Chapter 4), experimental games (Chapter 5) and scenario
interviews (Chapter 6), have the potential to inform the development of an agentbased model of smallholder land clearance. Despite the examples listed above of
experimental games being combined with agent-based models, efforts in this direction
have so far been limited and have not made use of the advances made in recent years
in taking games out of the lab into the field. In so doing, future research could
advance the field of land use multi-agent systems by modelling real world agents and
their environment and directly parameterising agent characteristics using the
behavioural preferences elicited through the use of experimental games and scenarios.
136
The creation of such a model would enable the investigation of the decision-making
processes underlying smallholder land clearance, with the associated potential for the
refinement of conservation interventions that this offers.
137
References
Adams, W. (2004) Against extinction. Earthscan, London.
Adams, W., Aveling, R., Brockington, D., Dickson, B., Elliot, J., Hutton, J., Roe, D.,
Vira, B. Wolmer, W. (2004) Biodiversity conservation and the eradication of
poverty. Science. 306 (5699), 1146–1149.
Adams, W., Brockington, D., Dyson, J. & Vira, B. (2003) Managing tragedies:
understanding conflict over common pool resources. Science. 302 (5652), 1915–
1916.
Adams, W. & Hulme, D. (2001) If community conservation is the answer in Africa,
what is the question? Oryx. 35 (3), 193–200.
ADB (2002) Indigenous peoples/ethnic minorities and poverty reduction, Cambodia.
Asian Development Bank, Manila.
Adger, W.N. (2006) Vulnerability. Global Environmental Change. 16 (3), 268–281.
ADHOC (2013) A turning point? Land, housing and natural resources rights in
Cambodia in 2012. Cambodian Human Rights and Development Association
(ADHOC), Phnom Penh.
Agrawal, A. (2014) Matching and mechanisms in protected area and poverty
alleviation research. Proceedings of the National Academy of Sciences of the
United States of America. 111 (11), 3909–3910.
Agrawal, A. & Gibson, C.C. (1999) Enchantment and Disenchantment: The Role of
Community in Natural Resource Conservation. World Development. 27 (4), 629–
649.
Akaike, H. (1974) A new look at the statistical model identification. IEEE
Transactions on Automatic Control. 19 (6), 716–723.
Alix-Garcia, J., De Janvry, A. & Sadoulet, E. (2008) The role of deforestation risk
and calibrated compensation in designing payments for environmental services.
Environment and Development Economics. 13 (3), 375–394.
Andam, K.S., Ferraro, P.J., Pfaff, A., Sanchez-Azofeifa, G.A. & Robalino, J.A.
(2008) Measuring the effectiveness of protected area networks in reducing
deforestation. Proceedings of the National Academy of Sciences of the United
States of America. 105 (42), 16089–16094.
Andam, K.S., Ferraro, P.J., Sims, K.R.E., Healy, A. & Holland, M.B. (2010)
Protected areas reduced poverty in Costa Rica and Thailand. Proceedings of the
National Academy of Sciences of the United States of America. 107 (22), 9996–
10001.
Angelsen, A. & Kaimowitz, D. (1999) Rethinking the Causes of Deforestation:
Lessons from Economic Models. The World Bank Research Observer. 14 (1),
73–98.
Arriagada, R.A., Ferraro, P.J., Sills, E.O., Pattanayak, S.K., et al. (2012) Do Payments
for Environmental Services Affect Forest Cover? A Farm-Level Evaluation from
Costa Rica. Land Economics. 88 (2), 382–399.
138
Asquith, N.M., Vargas, M.T. & Wunder, S. (2008) Selling two environmental
services: In-kind payments for bird habitat and watershed protection in Los
Negros, Bolivia. Ecological Economics. 65 (4), 675–684.
Baird, I. G. (2000) The Ethno-ecology, Land Use, and Livelihoods of the Brao-Kavet
Indigenous Peoples in Kok Lak Commune, Veun Sai District, Ratanakiri Province,
Northeast Cambodia. Ban Lung. The Non-Timber Forest Products Project,
Cambodia.
Baird, I.G. & Dearden, P. (2003) Biodiversity conservation and resource tenure
regimes: a case study from northeast Cambodia. Environmental management. 32
(5), 541–550.
Balmford, A. & Whitten, T. (2003) Who should pay for tropical conservation, and
how could the costs be met? Oryx. 37 (2), 238–250.
Barbier, E.B. (1997) The economic determinants of land degradation in developing
countries. Philosophical Transactions of the Royal Society B: Biological
Sciences. 352 (1356), 891–899.
Barnaud, C., Bousquet, F. & Trebuil, G. (2008) Multi-agent simulations to explore
rules for rural credit in a highland farming community of Northern Thailand.
Ecological Economics. 66 (4), 615-627.
Barrett, C.B. & Arcese, P. (1995) Are Integrated Conservation-Development Projects
(ICDPs) Sustainable? On the conservation of large mammals in sub-Saharan
Africa. World Development. 23 (7), 1073–1084.
Barrett, C.B. & Arcese, P. (1998) Wildlife Harvest in Integrated Conservation and
Development Projects: Linking Harvest to Household Demand, Agricultural
Production, and Environmental Shocks in the Serengeti. Land Economics. 74 (4),
449–465.
Barrett, C.B., Brandon, K., Gibson, C. & Gjertsen, H. (2001) Conserving Tropical
Biodiversity amid Weak Institutions. BioScience. 51 (6), 497.
Bates, D. & Maechler, M. (2012) lme4: Linear mixed-effects models using S4 classes.
R package version 0.999375-31. http://CRAN.R-project.org/package=lme4
Beckenkamp, M. & Ostman. A. (1999) Missing the Target? Sanctioning as an
Ambiguous Structural Solution. In: Foddy, M., Smithson, M., Scheneider, S. &
Hogg, M. (Eds.), Resolving Social Dilemmas: Dynamic, Structural, and
Intergroup Aspects. Psychology Press, London.
Berger, T. & Schreinemachers, P. (2006) Creating agents and landscapes for
multiagent systems from random samples. Ecology and society. 11 (2), 54-71.
Berkes, F. (2009) Community conserved areas: policy issues in historic and
contemporary context. Conservation Letters. 2 (1), 20–25.
Berkes, F. (2004) Rethinking Community-Based Conservation. Conservation Biology.
18 (3), 621–630.
Bernard, H.R., Killworth, P., Kronenfeld, D. & Sailer, L. (1984) The Problem of
Informant Accuracy: The Validity of Retrospective Data. Annual Review of
Anthropology. 13 (1), 495–517.
Biddulph, R. (2011) Tenure security interventions in Cambodia: testing Bebbington’s
139
approach to development geography. Geografiska Annaler: Series B, Human
Geography. 93 (3), 223-236.
Biel, A. & Thøgersen, J. (2007) Activation of social norms in social dilemmas: A
review of the evidence and reflections on the implications for environmental
behaviour. Journal of Economic Psychology. 28 (1), 93-112.
Blom, B., Sunderland, T. & Murdiyarso, D. (2010) Getting REDD to work locally:
lessons learned from integrated conservation and development projects.
Environmental Science & Policy. 13 (2), 164–172.
Bolker, B.M., Brooks, M.E., Clark, C.J., Geange, S.W., et al. (2009) Generalized
linear mixed models: a practical guide for ecology and evolution. Trends in
Ecology & Evolution. 24 (3), 127–135.
Börner, J. & Wunder, S. (2008) Paying for avoided deforestation in the Brazilian
Amazon: from cost assessment to scheme design. International Forestry Review.
10 (3), 496–511.
Bourdier, F. (2009) Ethnographie des populations indigènes du nord-est
Cambodgien: la montagne aux pierres précieuses (Ratanakiri). L'Harmattan,
Paris.
Bourgoin, J. (2012) Sharpening the understanding of socio-ecological landscapes in
participatory land-use planning. A case study in Lao PDR. Applied Geography.
34, 99–110.
Brandon, K. (1996) Ecotourism and conservation: A review of key issues. World
Bank, Environment Dept., Global Environment Division.
Brechin, S.R., Wilshusen, P.R., Fortwangler, C.L. & West, P.C. (2002) Beyond the
Square Wheel: Toward a More Comprehensive Understanding of Biodiversity
Conservation as Social and Political Process. Society & Natural Resources. 15
(1), 41–64.
Brockington, D. & Igoe, J. (2006) Eviction for Conservation: A Global Overview.
Conservation & Society, 4 (4), 24-70.
Brockington, D., Igoe, J. & Schmidt-Soltau, K. (2006) Conservation, Human Rights,
and Poverty Reduction. Conservation Biology. 20 (1), 250–252.
Brockington, D. & Schmidt-Soltau, K. (2004) The social and environmental impacts
of wilderness and development. Oryx. 38 (02), 140–142.
Brown, D., Seymour, F. & Peskett, L. (2008) How do we achieve REDD co-benefits
and avoid doing harm? In: Angelsen, A. (Ed.), Moving Ahead with REDD:
Issues, Options and Implications. Bogor: Center for International Forestry
Research, 107-118.
Brown, K. (2002) Innovations for conservation and development. The Geographical
Journal. 168 (1), 6–17.
Brown, K. (2003) Three challenges for a real people-centred conservation. Global
Ecology and Biogeography. 12 (2), 89–92.
Brown, T.C., Champ, P.A., Bishop, R.C., & McCollum, D.W. (1996). Which
response format reveals the truth about donations to a public good? Land
Economics. 72 (2), 152-166.
140
Bunnefeld, N., Hoshino, E. & Milner-Gulland, E.J. (2011) Management strategy
evaluation: a powerful tool for conservation? Trends in ecology & evolution. 26
(9), 441–447.
Burnham, K. & Anderson, D. (2002) Model selection and multi-model inference.
Springer, New York.
Cardenas, J.C. (2004) Norms from outside and from inside: an experimental analysis
on the governance of local ecosystems. Forest Policy and Economics. 6 (3-4),
229-241.
Cardenas, J.C. & Ostrom, E. (2004) What do people bring into the game?
Experiments in the field about cooperation in the commons. Agricultural Systems.
82 (3), 307–326.
Cardenas, J.C., Stranland, J. & Willis, C. 2000. Local environmental control and
institutional crowding-out. World Development. 28 (10), 1719-1733.
Carlsson, L. & Berkes, F. (2005) Co-management: concepts and methodological
implications. Journal of environmental management. 75 (1), 65–76.
Castella, J., Trung, T. & Boissau, S. 2005. Participatory simulation of land-use
changes in the northern mountains of Vietnam: The combined use of an agentbased model, a role-playing game, and a geographic information system. Ecology
and Society. 10 (1), 1-32.
Castillo, D. & Saysel, A.K. (2005) Simulation of common pool resource field
experiments: a behavioral model of collective action. Ecological Economics. 55
(3), 420–436.
CCBA (2013) Climate, Community & Biodiversity Standards Third Edition. CCBA,
Arlington.
CENTDOR (2012) Progress Activity Report: Basic Necessity Assessment, Seima
Biodiversity Conservation, Mondulkiri Province, Cambodia. CENTDOR, Phnom
Penh.
Cernea, M.M. & Schmidt-Soltau, K. (2006) Poverty Risks and National Parks: Policy
Issues in Conservation and Resettlement. World Development. 34 (10), 1808–
1830.
Chan, S. (2008) Impact of High Food Prices in Cambodia. Summary and Interim
Policy Brief No. 2. Cambodia Development Research Institute, Phnom Penh.
Chan, K.M.A., Pringle, R.M., Ranganathan, J., Boggs, C.L., Chan, Y.L., Ehrlich,
P.R., Haff, P.K., Al-Khafaji, K. & Macmynowski, D.P (2007) When agendas
collide: human welfare and biological conservation. Conservation Biology. 21
(1), 59–68.
Chomitz, K.M. (2007) At loggerheads? Agricultural expansion, poverty reduction,
and environment in the tropical forests, World Bank Publications, Washington,
D.C.
Chhatre, A. & Agrawal, A. (2008) Forest commons and local enforcement.
Proceedings of the National Academy of Sciences of the United States of
America. 105 (36), 13286–13291.
141
Chhatre, A. & Agrawal, A. (2009) Trade-offs and synergies between carbon storage
and livelihood benefits from forest commons. Proceedings of the National
Academy of Sciences of the United States of America. 106 (42), 17667–17670.
Cinner, J.E., Daw, T. & McClanahan, T.R. (2009) Socioeconomic factors that affect
artisanal fishers’ readiness to exit a declining fishery. Conservation Biology. 23
(1), 124–130.
Claridge, G., Veasna C-L. & In Van, C. (2005) The Effectiveness of Law
Enforcement Against Forest and Wildlife Crime: A Study of Enforcement
Disincentives and Other Relevant Factors in Southwestern Cambodia.
Conservation International, Phnom Penh.
Clements, T. (2010) Reduced Expectations: the political and institutional challenges
of REDD+. Oryx. 44 (03), 309–310.
Clements, T. & Milner-Gulland, E.J. (in press) The impacts of Payments for
Environmental Services and Protected Areas on local livelihoods and forest
conservation in northern Cambodia. Conservation Biology.
Clements, T., John, A., Nielsen, K., An D., Setha T. & Milner-Gulland, E.J. (2010)
Direct payments for biodiversity conservation: comparison of three schemes from
Cambodia. Ecological Economics. 69 (6), 1283-1291.
Clements, T.J., Soun S., An, D., Wilkie, D. & Milner-Gulland, E.J. (in press) Impacts
of forest conservation policies on local poverty and livelihoods in Cambodia.
World Development.
Colchester, M. (2004) Conservation policy and indigenous peoples. Environmental
Science & Policy. 7 (3), 145–153.
Cowling, R.M. (2014) Let’s get serious about human behavior and conservation.
Conservation Letters. [Online] Available from: doi:10.1111/conl.12106
[Accessed: 14 April 2014].
Cronkleton, P., Albornoz, M.A., Barnes, G., Evans, K. & Jong, W. (2010) Social
Geomatics: Participatory Forest Mapping to Mediate Resource Conflict in the
Bolivian Amazon. Human Ecology. 38 (1), 65–76.
Croson, R. (2005) The method of experimental economics. Internation Negotiation.
10 (1), 131-148.
Damania, R., Milner-Gulland, E.J. & Crookes, D.J. (2005) A bioeconomic analysis of
bushmeat hunting. Proceedings. Philosophical Transactions of the Royal Society
B: Biological Sciences. 272 (1560), 259–266.
Davies, R. (1997) Beyond wealth ranking: The democratic definition and
measurement of poverty. Briefing note. ODI workshop "Indicators of Poverty:
Operational Significance", London.
Daw, T., Brown, K., Rosendo, S. & Pomeroy, R. (2011) Applying the ecosystem
services concept to poverty alleviation: the need to disaggregate human wellbeing. Environmental Conservation. 38 (4), 370–379.
142
Deadman, P., Robinson, D., Moran, E., & Brondizio, E. (2004) Colonist household
decisionmaking and land-use change in the Amazon Rainforest: an agent-based
simulation. Environment and Planning B: Planning and Design. 31 (5), 693-709.
DeFries, R.S., Rudel, T., Uriarte, M. & Hansen, M. (2010) Deforestation driven by
urban population growth and agricultural trade in the twenty-first century. Nature
Geoscience. 3 (3), 178–181.
Drury, R., Homewood, K. & Randall, S. (2011) Less is more: the potential of
qualitative approaches in conservation research. Animal Conservation. 14 (1),
18–24.
Ducourtieux, O., Laffort, J.-R. & Sacklokham, S. 2005. Land Policy and Farming
Practices in Laos. Development and Change. 36 (3), 499–526.
Duflo, E., Kremer, M. & Robinson, J. (2008) How High Are Rates of Return to
Fertilizer? Evidence from Field Experiments in Kenya. The American Economic
Review. 98 (2), 482–488.
Ebeling, J. & Yasué, M. (2008) Generating carbon finance through avoided
deforestation and its potential to create climatic, conservation and human
development benefits. Philosophical transactions of the Royal Society of London:
Biological sciences. 363 (1498), 1917–1924.
Engel, S., Pagiola, S. & Wunder, S. (2008) Designing payments for environmental
services in theory and practice: An overview of the issues. Ecological Economics.
65 (4), 663–674.
Evans, T.D. (2007) A survey of communities in and around the Seima Biodiversity
Conservation Area in 2006. Wildlife Conservation Society, Phnom Penh.
Evans, T.D., Hout, P., Phet, P. & Hang M. (2003) A study of resin-tapping and
livelihoods in southern Mondulkiri, Cambodia, with implications for conservation
and forest managements. Wildlife Conservation Society, Phnom Penh.
Evans, T.D., O’Kelly, H.J., Men S., Nut M.H., Phet P. Som P. & Pollard, E. (2012a)
Seima Protection Forest, Cambodia. In: Sunderland, T.C., Sayer, J. & Hoang,
M.H. [Eds.], Evidence-based conservation: lessons from the Lower Mekong.
Routledge, Oxon.
Evans, T.D., Arpels, M. & Clements, T. (2012b) Pilot REDD Activities in Cambodia
are Expected to Improve Access to Forest Resource Use Rights and Land Tenure
for Local Communities. In: Naughton-Treves, L. and Day, C. [Eds.], Lessons
about Land Tenure, Forest Governance and REDD+. Case Studies from Africa,
Asia and Latin America. UW-Madison Land Tenure Center, Madison.
Evans, T.P., Sun, W. & Kelley, H. (2006) Spatially explicit experiments for the
exploration of land‐ use decision‐ making dynamics. International Journal of
Geographical Information Science. 20 (9), 1013-1037.
FA (2010) REDD+ roadmap in Cambodia (REDD+ readiness plan in Cambodia).
Forestry Administration, Phnom Penh.
(Fairhead, J., Leach, M. & Scoones, I. (2012) Green Grabbing: a new appropriation of
nature? Journal of Peasant Studies. 39 (2), 237–261.
FAO (2010) Global Forest Resources Assessment 2010. FAO Forestry Paper No 163.
FAO, Rome.
143
Ferraro, P.J. (2011) The future of payments for environmental services. Conservation
Biology. 25 (6), 1134–1138.
Ferraro, P.J. & Kiss, A. (2002) Direct payments to conserve biodiversity. Science. 298
(5599), 1718–1719.
Ferraro, P.J. & Pattanayak, S.K. (2006) Money for nothing? A call for empirical
evaluation of biodiversity conservation investments. PLoS biology. 4 (4), e105.
Fischbacher, U., Gachter, S. & Fehr, E. (2001) Are people conditionally cooperative?
Evidence from a public goods experiment. Economics Letters. 71 (3), 397–404.
Foster, V., Bateman, I.J. & Harley, D. (1997) Real and hypothetical willingness to
pay for environmental preservation: a non-experimental comparison. Journal of
Agricultural Economics. 48 (1-3), 123–137.
Fox, J., Troung, D., Rambo, A., Tuyen, N. & Leisz, S. (2000) Shifting cultivation: a
new old paradigm for managing tropical forests. Bioscience. 50 (6), 521-528.
Fox, J., Mcmahon, D., Poffenberger, M. & Vogler, J. (2008) Land for my
grandchildren: land use and tenure change in Ratanakiri 1989-2007. Community
Forestry International and East-West Center.
Frey, B. & Jegen, R. (2001) Motivation crowding theory. Journal of Economic
Surveys. 15 (5), 589–611.
Frey, B. & Oberhozer-Gee, F. (1997) The cost of price incentives: an empirical
analysis of motivation crowding-out. The American Economic Review. 87, 746755.
Game, E.T., Meijaard, E., Sheil, D. & McDonald-Madden, E. (in press) Conservation
in a wicked complex world; challenges and solutions. Conservation Letters.
Available from: doi:10.1111/conl.12050 [Accessed: 29 April 2014].
Geist, H. & Lambin, E. (2002) Proximate Causes and Underlying Driving Forces of
Tropical Deforestation. BioScience. 52 (2), 143–150.
Ghimire, K.B. & Pimbert, M.P. (1997) Social change and conservation: an overview
of issues and concepts. In: Ghimire, K.B., Ghimire, K.B. & Pimbert, M.P. (Eds.),
Social change and conservation: environmental politics and impacts of national
parks and protected areas. Earthscan, Oxford, pp. 1-45.
Gibson, C.C., Williams, J.T. & Ostrom, E. (2005) Local Enforcement and Better
Forests. World Development. 33 (2), 273–284.
Gintis, H. (2000) Beyond Homo economicus: evidence from experimental economics.
Ecological economics. 35 (3), 311–322.
Grimsditch, M. & Henderson, N. (2009) Untitled: Tenure Insecurity and Inequity in
the Cambodian Land Sector. Bridges Across Borders Southeast Asia, Centre on
Housing Rights and Evictions, Jesuit Refugee Service, Phnom Penh.
Gordon, T. J. (1992) The Methods of Futures Research. Annals AAPSS. 522 (1), 2535.
Hansen, M.C., Potapov, P. V, Moore, R., Hancher, M., Turubanova, S.A., Tyukavina,
A., Thau, D., Stehman, S.V., Goetz, S.J., Loveland, T.R., Kommareddy, A.,
Egorov, A., Chini, L., Justice, C.O. Townshend, J.R.G. (2013) High-resolution
global maps of 21st-century forest cover change. Science. 342 (6160), 850–853.
144
Hardin, G. (1968) The tragedy of the commons. Science. 162 (5364), 1243-1248.
Hayo, B. & Vollan, B. (2012) Group interaction, heterogeneity, rules, and cooperative behaviour: Evidence from a common-pool resource experiment in
South Africa and Namibia. Journal of Economic Behavior & Organization. 81
(1), 9–28.
Heckbert, S., Baynes, T. & Reeson, A. (2010) Agent‐ based modeling in ecological
economics. Annals of the New York Academy of Sciences. 1185 (1), 39-53.
Henrich, J., Boyd, R., Bowles, S., Camerer, C., Fehr, E., Gintis, H., McElreath, R.,
Alvard, M., Barr, A., Ensminger, J., Henrich, N., Hill, K., Gil-White, F., Gurven,
M., Marlowe, F., Patton, J. & Tracer, D. (2005) “Economic man” in crosscultural perspective: behavioural experiments in 15 small-scale societies.
Behavioural and Brain Sciences. 28 (6), 795-815.
Hirsch, P.D., Adams, W.M., Brosius, J.P., Zia, A., Bariola, N., & Dammert, J.L.
(2011) Acknowledging conservation trade-offs and embracing complexity.
Conservation Biology. 25 (2), 259–264.
Huigen, M.G.A., 2004. First principles of the MameLuke multi-actor modelling
framework for land use change, illustrated with a Philippine case study. Journal
of Environmental Management, 72(1-2), 5-21.
Keane, A., Jones, J.P.G., Edwards-Jones, G. & Milner-Gulland, E.J. (2008) The
sleeping policeman: understanding issues of enforcement and compliance in
conservation. Animal Conservation. 11 (2), 75–82.
Kellert, S.R., Mehta, J.N., Ebbin, S.A. & Lichtenfeld, L.L. (2000) Community natural
resource management: promise, rhetoric, and reality. Society and Natural
Resources. 13 (8), 705-715.
Kheang U. (2006) State, Society and Democratic Consolidation: The Case of
Cambodia. Pacific Affairs. 79 (2), 225-245.
Kocher, M.G. & Sutter, M. (2005) The Decision Maker Matters: Individual Versus
Group Behaviour in Experimental Beauty-Contest Games. The Economic
Journal. 115 (500), 200–223.
Kuoy, B. (2013) Ethics and Cambodian worldviews on nature. Ethics in Science and
Environmental Politics. 13 (1), 49-57.
Janssen, M.A., 2005. Agent-based modeling. In: Proops, J. & Safonov, P. [Eds.],
Modeling in Ecological Economics. Edward Elgar Publising, Cheltnam. 155-172.
Jones, J.P.G., Andriamarovololona, M.M. & Hockley, N. (2008) The Importance of
Taboos and Social Norms to Conservation in Madagascar. Conservation Biology.
22 (4), 976–986.
Lambin, E.F., Turner, B.L., Geist, H.J., Agbola, S.B., Angelsen, A., Bruce, J.W.,
Coomes, O.T., Dirzo, R., Fischer, G., Folke, C., George, P.S., Homewood, K.,
Imbernon, J., Leemans, R., Li, X., Moran, E.F., Mortimore, M., Ramakrishnan,
P.S., Richards, J.S., Skånes, H., Steffen, W., Stone, G.D., Svedin, U., Veldkamp,
T.A., Vogel, C. & Xu, J. (2001) The causes of land-use and land-cover change:
moving beyond the myths. Global Environmental Change. 11 (4), 261–269.
Larson, A.M. (2011) Forest tenure reform in the age of climate change: Lessons for
REDD+. Global Environmental Change. 21 (2), 540–549.
145
Laurance, W. (1999) Reflections on the tropical deforestation crisis. Biological
Conservation. 91 (2-3), 109–117.
Lee, E. & Mahanty, S. (2009) Payments for environmental services and poverty
reduction: opportunities and risks. RECOFTC, Bangkok.
Lestrelin, G. (2010) Land degradation in the Lao PDR: Discourses and policy. Land
Use Policy. 27 (2), 424–439.
Lestrelin, G., Bourgoin, J., Bouahom, B. & Castella, J.-C. (2011) Measuring
participation: Case studies on village land use planning in northern Lao PDR.
Applied Geography. 31 (3), 950–958.
Levitt, S. D., & List, J. A. (2007). What do laboratory experiments measuring social
preferences reveal about the real world? The journal of economic perspectives. 21
(2), 153-174.
Loomis, J., Brown, T., Lucero, B., & Peterson, G. (1996) Improving validity
experiments of contingent valuation methods: Results of efforts to reduce the
disparity of hypothetical and actual willingness to pay. Land Economics. Nov96.
72 (4).
Lopez, M.C, Murphy, J.J., Spraggon, J.M. & Stranlund, J.K. (2012) Comparing the
effectiveness of regulation and pro-social emotions to enhance cooperation:
experimental evidence from fishing communities in Colombia. Economic Inquiry.
50, 131-142.
Mansuri, G. & Rao, V. (2004) Community-Based and -Driven Development: A
Critical Review. The World Bank Research Observer. 19 (1), 1–39.
Margoluis, R., & Salafsky, N. (1998). Measures of success: designing, managing, and
monitoring conservation and development projects. Island Press, Washington,
DC.
Martin, D. (2013) Measuring the effectiveness of incentives: can experimental games
really give us the answer? MSc Dissertation. Imperial College, London.
Matthews, R.B., Gilbert, N.G., Roach, A., Polhill, J.G., & Gotts, N.M. (2007) Agentbased land-use models: a review of applications. Landscape Ecology. 22 (10),
1447-1459.
McAndrew, J.P. (2009) Access to natural resources: case studies of Cambodian hill
tribes. In: Perera, J. (Ed.), Land and cultural survival: The communal land rights
of indigenous peoples in Asia. Asia Development Bank, Manila.
McAndrew, J.P., Mam S., Hong K. & Ly B. (2003) Indigenous adpatation to a
decline in natural resources: the experience of two Phong communities in
Northeast Cambodia. CIDSE, Phnom Penh.
McCall, M. K. & Minang, P. A. (2005) Assessing Participatory GIS for CommunityBased Natural Resource Management: Claiming Community Forests in
Cameroon. The Geographical Journal. 171, (4), 340-356.
McKennney, B., Chea, Y., Tola, P. & Evans, T. (2004) Focusing on Cambodia’s
High Value Forests: Livelihoods and Management. Cambodia Development
Research Institute & Wildlife Conservation Society, Phnom Penh.
McMahon, D. (2008) Assessment of Community Forestry Sites and Migration
146
Patterns in the Oddar Mean Chey Province, Cambodia. Community Forestry
International, Phnom Penh.
McShane, T.O., Hirsch, P.D., Trung, T.C., Songorwa, A.N., Kinzig, A., Monteferri,
B., Mutekanga, D., Thang, H.V., Dammert, J.L., Pulgar-Vidal, M., WelchDivine, M., Brosius, P.J., Coppolillo, P. & O’Connor, S. (2011) Hard choices:
Making trade-offs between biodiversity conservation and human well-being.
Biological Conservation. 144 (3), 966–972.
McShane, T.O. & Newby, S.A. (2004.) Expecting the unattainable: the assumptions
behind ICDPs. In: McShane, T. O., Wells, M.P. (Eds.), Getting Biodiversity
Projects to Work. Columbia University Press, New York, pp. 49–74.
Miles, L. & Kapos, V. (2008) Reducing Greenhouse Gas Emissions from
Deforestation and Forest Degradation: Global Land-Use Implications. Science.
320 (5882), 1454 -1455.
Milne, S. (2013) Under the leopard’s skin: Land commodification and the dilemmas
of indigenous communal title in upland Cambodia. Asia Pacific Viewpoint. 54 (3),
323-339.
Milne, S., Evans, T. & Travers, H. (2012) Policy options for REDD benefit-sharing at
the site level in Cambodia: a conceptual overview and early recommendations
for the Seima Protection Forest Demonstration Project site. Wildlife
Conservation Society, Phnom Penh.
Milne, S. & Niesten, E. (2009) Direct payments for biodiversity conservation in
developing countries: practical insights for design and implementation. Oryx. 43
(4), 530-541.
Milner-Gulland, E.J. (2012) Interactions between human behaviour and ecological
systems. Philosophical transactions of the Royal Society of London. Series B,
Biological sciences. 367 (1586), 270–278.
Milner-Gulland, E.J. & Bennett, E.L. (2003) Wild meat: the bigger picture. Trends in
Ecology & Evolution. 18 (7), 351–357.
Milner-Gulland, E.J., McGregor, J.A., Agarwala, M., Atkinson, G., Bevan, P.,
Clements, T., Daw, T., Homewood, K., Kumpel, N., Lewis, J., Mourato, S.,
Palmer Fry, B., Redshaw, M., Rowcliffe, J.M., Suon, S., Wallace, G.,
Washington, H. & Wilkie, D. (2014) Accounting for the Impact of Conservation
on Human Well-Being. Conservation Biology. [Online; accessed: 24 March
2014].
Mullahy, J. 1986. Specification and testing of some modified count data models.
Journal of Econometrics. 33 (3), 341–365.
Nakagawa, S. & Schielzeth, H. (2013) A general and simple method for obtaining R 2
from generalized linear mixed-effects models. Methods in Ecology and
Evolution. 4 (2), 133–142.
Narloch, U., Pascual, U. & Drucker, A.G. (2012) Collective Action Dynamics under
External Rewards: Experimental Insights from Andean Farming Communities.
World Development. 40 (10), 2107–2096.
147
Naughton-Treves, L., Alix-Garcia, J. & Chapman, C.A. (2011) Lessons about parks
and poverty from a decade of forest loss and economic growth around Kibale
National Park, Uganda. Proceedings of the National Academy of Sciences of the
United States of America. 108 (34), 13919–13924.
Neill, H. R., Cummings, R. G., Ganderton, P. T., Harrison, G. W., & McGuckin, T.
(1994). Hypothetical surveys and real economic commitments. Land Economics.
70 (2), 145-154.
North, D.C. (1990) Institutions, institutional change, and economic performance,
Cambridge University Press.
Nuno, A. (2013) Managing social-ecological systems under uncertainty: implications
for conservation. PhD Thesis. Imperial College, London.
Oberndorf, R.B. (2005) Legal analysis of forest and land laws in Cambodia.
Community Forestry International, Phnom Penh.
Ostmann, A. (1998) External control may destroy the commons. Rationality and
Society. 10 (1), 103–122.
Ostrom, E. (1990) Governing the commons: the evolution of institutions for collective
action. Cambridge University Press, Cambridge.
Ostrom, E. (2000) Collective action and evolution of social norms. Journal of
Economic Perspectives. 14 (3), 137-158.
Ostrom, E., Gardner, R. & Walker, J. (1994) Rules, games, and common-pool
resources. University of Michigan Press, Ann Arbor.
Ostrom, E., Walker, J., Gardner, R. (1992) Covenants with and without a sword: selfgovernance is possible. The American Political Science Review. 86 (2), 404–417.
Pagiola, S., Agostini, P., Gobbi, J., de Haan, C., Ibrahim, M., Murgueitio, E., Ramírez,
E., Rosales, M. & Ruíz, J. P. (2004) Paying for biodiversity conservation services
in agricultural landscapes. The World Bank Environment Department,
Washington, D.C.
Pascual, U., Muradian, R., Rodríguez, L.C. & Duraiappah, A. (2010) Exploring the
links between equity and efficiency in payments for environmental services: A
conceptual approach. Ecological Economics. 69 (6), 1237–1244.
Pendleton, L.H. & Howe, E.L. (2002) Market Integration, Development, and
Smallholder Forest Clearance. Land Economics. 78 (1), 1–19.
Persha, L., Agrawal, A. & Chhatre, A. (2011) Social and ecological synergy: local
rulemaking, forest livelihoods, and biodiversity conservation. Science. 331
(6024), 1606–1608.
Petheram, L. & Campbell, B.M. (2010) Listening to locals on payments for
environmental services. Journal of Environmental Management. 91 (5), 1139–
1149.
Phelps, J., Webb, E.L. & Agrawal, A. (2010) Land use. Does REDD+ threaten to
recentralize forest governance? Science. 328 (5976), 312–313.
Pollard, E. & Evans, T. (2008) A survey of communities in and around the Seima
Biodiversity Conservation Area in 2008. Wildlife Conservation Society, Phnom
Penh.
148
Proctor, W., Koellner, T. and Lukasiewicz, A. (2009) Equity Considerations and
Markets for Ecosystem Services. In: Kumar, P. and Muradian, R. (Eds.),
Payments for Ecosystem Services. Oxford University Press, New Delhi. pp. 1743.
Pullin, A.S. & Knight, T.M. (2009) Doing more good than harm – Building an
evidence-base for conservation and environmental management. Biological
Conservation. 142 (5), 931–934.
Pullin, A.S. & Knight, T.M. (2003) Support for decision making in conservation
practice: an evidence-based approach. Journal for Nature Conservation. 11 (2),
83–90.
Reeson, A.F. & Tisdell, J.G. (2008) Institutions, motivations and public goods: An
experimental test of motivational crowding. Journal of Economic Behavior &
Organization. 68 (1), 273–281.
Reichhuber, A., Camacho, E. & Requate, T. (2009) A framed field experiment on
collective enforcement mechanisms with Ethiopian farmers. Environment and
Development Economics. 14 (5), 641-663.
Richards, M., Panfil, S.N. (2011) Social and Biodiversity Impact Assessment (SBIA)
Manual for REDD+ Projects: Part 1 – Core Guidance for Project Proponents.
Version 2. Climate, Community & Biodiversity Alliance, Forest Trends, Fauna &
Flora International, and Rainforest Alliance, Washington D.C.
Robinson, D.T., Brown, D.G., Parker, D.C., Schreinemachers, P., Janssen, M.A.,
Huigen, M., Wittmer, H., Gotts, N., Promburom, P., Irwin, E., Berger, T.,
Gatzweiler, F. & Barnaud, C. (2007) Comparison of empirical methods for
building agent-based models in land use science. Journal of Land Use Science. 2
(1), 31–55.
Rock, F. (2001) Participatory land use planning in rural Cambodia: a manual for
government staff and development workers. Ministry of Land Management,
Urban Planning and Construction, Phnom Penh.
Roe, D. & Elliott, J. (2006) Pro-poor conservation: The elusive win-win for
conservation and poverty reduction? Policy Matters. 14, 53-63.
Rustagi, D., Engel, S. & Kosfeld, M. (2010) Conditional Cooperation and Costly
Monitoring Explain Success in Forest Commons Management. Science. 330
(6006), 961–965.
Salafsky, N., Margoluis, R., Redford, K.H. & Robinson, J.G. (2002) Improving the
Practice of Conservation: a Conceptual Framework and Research Agenda for
Conservation Science. Conservation Biology. 16 (6), 1469–1479.
Salafsky, N. & Wollenberg, E. (2000) Linking Livelihoods and Conservation: A
Conceptual Framework and Scale for Assessing the Integration of Human Needs
and Biodiversity. World Development. 28 (8), 1421–1438.
Scott, J.C. (2009). The art of not being governed: An anarchist history of upland
Southeast Asia. Yale University Press, New Haven.
Scrieciu, S.S. (2007) Can Economic Causes of Tropical Deforestation Be Identified at
a Global Level? Ecological Economics. 62(3/4), 603-12.
149
Sethi, R. & Somanathan, E. (1996) The evolution of social norms in common
property resource use. American Economic Review. 86 (4), 766-788.
Shahabuddin, G. & Rao, M. (2010) Do community-conserved areas effectively
conserve biological diversity? Global insights and the Indian context. Biological
Conservation. 143 (12), 2926–2936.
Simbolon, I. (2009) Law reforms and recognition of indigenous peoples' communal
rights in Cambodia. In: Perera, J. (Ed.), Land and cultural survival: The
communal land rights of indigenous peoples in Asia. Asia Development Bank,
Manila.
Simon, H., 1972. Theories of Bounded Rationality: A volume in honor of Jacob
Marschak. In: McGuire, C. & Radner, R. [Eds.], Decision and Organisation.
North-Holland Publishing Company, Amsterdam.
Singleton, S. (2000) Cooperation or capture? The paradox of co-management and
community participation in natural resource management and environmental
policy-making. Environmental Politics. 9 (2), 1–21.
Skutsch, M., Vickers, B., Georgiadou, Y. & McCall, M. (2011) Alternative models
for carbon payments to communities under REDD+: A comparison using the
Polis model of actor inducements. Environmental Science & Policy. 14 (2), 140–
151.
Sommerville, M., Jones, J.P.G., Rahajaharison, M. & Milner-Gulland, E.J. (2010) The
role of fairness and benefit distribution in community-based Payment for
Environmental Services interventions: A case study from Menabe, Madagascar.
Ecological Economics. 69 (6), 1262–1271.
Stem, C., Margoulis, R., Salafsky, N. & Brown, M. (2005) Monitoring and Evaluation
in Conservation: a Review of Trends and Approaches. Conservation Biology. 19
(2), 295–309.
Stickler, C.M., Nepstad, D.C., Coe, M.T., McGrath, D.G., Rodrigues, H.O., Walker,
W.S., Soares-Filho, B.S. & Davidson, E.A. (2009) The potential ecological costs
and cobenefits of REDD: a critical review and case study from the Amazon
region. Global Change Biology. 15 (12), 2803–2824.
Sunderlin, W., Hatcher, J. & Liddle, M. (2008) From Exclusion to Ownership?
Challenges and Opportunities in Advancing Forest Tenure Reform. Rights and
Resource Initiative, Washington D.C.
Sunderlin, W.D., Angelsen, A., Belcher, B., Burgers, P., Nasi, R., Santoso, L., &
Wunder, S. (2005) Livelihoods, forests, and conservation in developing
countries: An Overview. World Development. 33 (9), 1383-1402.
Sunderland, T.C.H., Ehringhaus, C. & Campbell, B.M. (2008) Conservation and
development in tropical forest landscapes: a time to face the trade-offs?
Environmental Conservation. 34 (4), 276–279.
Sutherland, W.J., Pullin, A.S., Dolman, P.M. & Knight, T.M. (2004) The need for
evidence-based conservation. Trends in ecology & evolution. 19 (6), 305–308.
Sutter, M., Haigner, S. & Kocher, M.G. (2010) Choosing the Carrot or the Stick?
Endogenous Institutional Choice in Social Dilemma Situations. Review of
Economic Studies. 77 (4), 1540–1566.
150
Terborgh, J. (2004) Requiem for nature. Island Press, Washington, DC.
Travers, H., Clements, T., Keane, A. & Milner-Gulland, E.J. (2011) Incentives for
cooperation: The effects of institutional controls on common pool resource
extraction in Cambodia. Ecological Economics. 71,151–161.
Travers, H. & Evans, T. (2013) Development of a social impacts monitoring system
for the Seima Core Protection Forest REDD+ Demonstration Site. Wildlife
Conservation Society, Phnom Penh.
UNDP (2010) Phnong ethnicity: documentation of customary rules of indigenous
people in Pu Char village, Sre Preah Commune, Keo Seima District, Mondulkiri
Province. UNDP, Phnom Penh.
UNDP (2011) Cambodia Human Development Report 2011. Building Resilience: The
Future of Rural Livelihoods in the Face of Climate Change. UNDP, Phnom Penh.
UNFCCC (2010) Report of the Conference of the Parties on its sixteenth session, held
in Cancun from 29 November to 10 December 2010. UNFCCC, Bonn.
UNFCCC (2011) Report of the Conference of the Parties on its seventeenth session,
held in Durban from 28 November to 11 December 2011. UNFCCC, Bonn.
UNFCCC (2013) Report of the Conference of the Parties on its seventeenth session,
held in Warsaw from 11 to 12 November 2013. UNFCCC, Bonn.
UNOHCHR (2011) 3 indigenous communities to receive communal land title. United
Nations Office of the High Commissioner for Human Rights in Cambodia,
Phnom Penh.
Velez, M.A., Murphy, J.J. & Stranlund, J.K. (2006) Centralized and
decentralize management of local common pool resources in the developing
world: experimental evidence from fishing communities in Colombia. Economic
Inquiry. 48, 254-265.
Venkatachalam, L. (2004) The contingent valuation method: a review. Environmental
Impact Assessment Review. 24 (1), 89–124.
Vollan, B. (2008) Socio-ecological explanations for crowding-out effects from
economic field experiments in southern Africa. Ecological Economics. 67 (4),
560-573.
Vollan, B. & Ostrom, E. (2010) Cooperation and the Commons. Science. 330 (6006),
923-924.
Voors, M., Turley, T., Kontoleon, A., Bulte, E. & List, J.A. (2012) Exploring whether
behavior in context-free experiments is predictive of behavior in the field:
Evidence from lab and field experiments in rural Sierra Leone. Economics
Letters. 114 (3), 308–311.
Walston, J., Davidson, P. & Men S. (2001) A wildlife survey in southern Modulkiri
Province, Cambodia. Wildlife Conservation Society, Phnom Penh.
Waylen, K.A., Fischer, A., McGowan, P.J.K. & Milner-Gulland, E.J. (2012)
Interactions Between a Collectivist Culture and Buddhist Teachings Influence
Environmental Concerns and Behaviors in the Republic of Kalmykia, Russia.
Society & Natural Resources. 25 (11), 1118–1133.
151
Waylen, K., Fischer, A., McGowan, P. & Milner-Gulland, E.J. (2013) Deconstructing
community for conservation: why simple assumptions are not sufficient. Human
Ecology. 41, 575-585.
WCS (2013) Reduced Emissions from Deforestation and Degradation in Seima
Protection Forest. Wildlife Conservation Society, Phnom Penh.
Wells, M. (1992) Biodiversity conservation, affluence and poverty: mismatched costs
and benefits and efforts to remedy them. Ambio. 21, 237-242.
Wells, M.P. & McShane, T.O. (2004) Integrating Protected Area Management with
Local Needs and Aspirations. Ambio. 33 (8), 513–519.
Wells, M. P. & Brandon, K. E. (1993) The principles and practice of buffer zones and
local participation in biodiversity conservation. Ambio. 22 (2/3), 157-162.
Wilkie, D.S., Carpenter, J.F. & Zhang, Q. (2001) The under-financing of protected
areas in the Congo Basin: so many parks and so little willingness-to-pay.
Biodiversity & Conservation. 10 (5), 691–709.
Wilkie, D.S., Starkey, M., Abernethy, K., Effa, E.N., Telfer, P. & Godoy, R. (2005)
Role of Prices and Wealth in Consumer Demand for Bushmeat in Gabon, Central
Africa. Conservation Biology. 19 (1), 268–274.
Wooldridge, M.J. (2002) An introduction to multiagent systems. John Wiley and Sons,
Chichester.
World Bank (2009) Poverty profile and trends in Cambodia, 2007. Findings from the
Cambodia Socio-Economic Survey. Report No. 486188KH. Poverty Reduction
and Economic Management Sector Unit, East Asia and Pacific Region. World
Bank, Washington D.C.
Wunder, S. (2001) Poverty Alleviation and Tropical Forests—What Scope for
Synergies? World Development. 29 (11), 1817–1833.
Wunder, S. (2005) Payments for environmental services: some nuts and bolts. CIFOR
Occasional Paper No. 42. CIFOR, Bogor.
Wunder, S. (2007) The efficiency of payments for environmental services in tropical
conservation. Conservation Biology. 21 (1), 48–58.
Zuur, A., Ieno, E., Walker, N., Saveliev, A., Smith, G. 2009. Mixed effects models
and extensions in ecology with R. Springer, New York.
152
Appendix A
8. Supplementary material for Chapter 3
Table A.1: Items included in the household poverty score, with the weighting used for each item,
percentage of households that reported owning or having access to each item and whether each
item was included in calculating the reduced poverty score used for comparison between 2006/7
and 2012.
Basic necessity survey item
three meals per day
VCD player
mosquito net for all household members
able to attend all weddings invited to
car battery (more than 40 A)
two draft animals
120L water jar
access to electricity (public supply or generator)
thick blanket for all household members
long knife
motorised tiller (koyun)
axe
hand pump well at home
home toilet
wooden wardrobe
access to car taxi service
motorbike
zinc/fibro/tile roof
wooden walls
television
mobile phone
homestead land at least 0.5ha
at least 3 ha chamkar
water supply to home
able to send children to school to grade 9
ability to attend all traditional ceremonies
two sets of clothes for every household member
synthetic tent for sleeping in the forest
hammock with mosquito net
ox cart
strimmer
kettle
153
Weights
Own/have
access to
Reduced
score
100.0%
63.7%
99.5%
95.8%
86.8%
89.1%
98.2%
95.2%
91.6%
98.6%
78.1%
98.7%
95.7%
93.7%
78.1%
85.4%
98.9%
99.8%
99.7%
81.0%
90.2%
96.3%
99.7%
98.9%
96.3%
98.6%
95.3%
93.6%
96.6%
75.1%
84.6%
93.4%
79.4%
30.1%
84.7%
64.6%
29.7%
25.7%
54.2%
12.4%
21.1%
92.1%
12.2%
90.5%
12.9%
9.2%
10.3%
19.3%
81.7%
84.7%
69.0%
30.1%
72.0%
26.0%
38.9%
0.3%
11.7%
72.2%
28.0%
41.2%
35.5%
20.4%
29.7%
86.5%
no
no
no
no
yes
yes
yes
no
no
no
no
no
no
yes
no
no
yes
yes
yes
yes
yes
yes
yes
no
no
no
no
no
no
no
no
no
Table A.2: Variables used in statistical modelling in Chapter 3.
Variable
poverty_score
adj_score
resin
trees
log_area
ntfp_collect
wild_meals
widow
dependent
landless
age
yrs_village
gender
indigenous
AME
hh_size
education
km
position
labour
shop
employed
service
debt
hh_labour
labour_avail
security
core
year
village
Data type
continuous
continuous
binary
categorical
continuous
binary
count
binary
binary
binary
continuous
continuous
categorical
binary
continuous
continuous
continuous
continuous
binary
binary
binary
binary
binary
binary
binary
binary
binary
binary
categorical
categorical
Variable type
response/explanatory
response/explanatory
response/explanatory
response/explanatory
response/explanatory
response/explanatory
response
explanatory
explanatory
explanatory
explanatory
explanatory
explanatory
explanatory
explanatory
explanatory
explanatory
explanatory
explanatory
explanatory
explanatory
explanatory
explanatory
explanatory
explanatory
explanatory
explanatory
explanatory
explanatory
random
Description
poverty score of respondent household
reduced poverty score of respondent household
whether or not the respondent household owned any resin trees
quantity of resin trees owned (0-50, 51-150,>150)
ln(total area of land claimed by respondent household)
whether or not the respondent household collected non-liquid resin NTFPs
number of meals eaten in previous week containing wild meat
whether or not the respondent household was widow headed
whether or not the respondent household consisted totally of dependents (<15s or >65s)
whether or not the respondent household owned any productive land
age of household head
number of years respondent had lived in village
gender of respondent
whether or not the the head of the respondent household was an IP
household adult male equivalence calculated using Barba and Cabrera (2008)
total number of members in respondent household
number of years spent in education by the household head
distance to nearest all day market [km]
whether or not the respondent household held a formal position of authority
whether or not the respondent household was involved in wage labour
whether or not the respondent household operated a shop
whether or not a member of respondent’s household was employed
whether or not the respondent’s household operated a village service
whether or not the respondent household was in debt
whether or not the respondent household felt they had sufficient labour within the household
whether or not the respondent household felt their was sufficient labour available
whether or not the respondent household felt their natural resources were secure
whether or not the village was located in the core zone
year of survey
village in which respondent household lived
154
Appendix B
9. Supplementary material for Chapter 4
Calculation of adjusted household poverty score
The basic necessity survey provides a locally defined poverty index, which can be
used to score households sampled in the area of the survey (Davies, 1997). In Chapter
4, the results of the large scale socio-economic survey described in Chapter 3, which
was undertaken in Seima Protection Forest (SPF) at the time of the measurement of
land use in the two villages and contained a basic necessity component, were used.
The basic necessity component of this survey consisted of a series of 40 household
items, assets and services. Respondent households were asked whether or not they
owned the items on the list or, in the case of services, whether they had access to
those services. For example one of the questions on the list asked whether the
respondent household had access to a car-taxi service to and from the district capital
or nearest market town. Respondents were also asked whether or not they believed
that each item on the list was a basic necessity, which was defined to them (following
the adapted definition used by Clements et al. (in press)) as the “minimum
requirement for living that all households of the community should have and no-one
should not have.”
The list was created at a one-day workshop, held at the SPF headquarters of the
Forestry Administration in April 2012, and to which representatives of different
livelihood zones within SPF and different wealth bands were invited. These
representatives were selected by FA or WCS staff, who were able to identify suitable
candidates for participation. Over the course of the one-day workshop, participants
were introduced to the concept of basic necessities and compiled the list of items. The
list was proportionally populated with (i) items that everyone thought were basic
necessities and that everyone would have (or would have access to) (ii) items that
everyone thought were basic necessities, and around half of all people would have,
but would get as they became richer and services improved (iii) items that everyone
thought were basic necessities, but only some people would have, but might get as
they became richer and services improved (iv) items <50% of people thought were
155
basic necessities. The list was populated in this way to ensure that there was a good
level of differentiation between households, that respondents were made to think
whether or not each item really met the definition of a basic necessity and that the
same list of items could be used again in the future after incomes were anticipated to
rise.
Once the survey instrument had been compiled, it was piloted in three villages to
ensure that the items fell roughly in proportion to the different categories identified.
No items were changed following the pilot and the survey was conducted in a further
17 villages across the SPF locality, covering a total of 622 households. A household
poverty score was calculated for each household using only those items that over 50%
of sampled households thought were basic necessities (in this case 31 items). Scores
were weighted by the proportion of the sample that thought each item met the
definition of a basic necessity and these scores were summed over the full list of 31
items for each household to produce a household poverty score. For example, if 95%
of the sample thought that a sharp knife was a basic necessity, the score for owning a
sharp knife would be 0.95.
In this analysis, asset data was collectedfor a smaller list of assets and then used these
assets to calculate an adjusted poverty score based on the coefficient estimates of a
LMM, with household poverty score as the response variable. This model was run for
the sample of 622 households that participated in the full basic necessity survey
(Table B.1). For the purposes of this analysis, the original household poverty scores
were recalculated after removing those items that corresponded to the items that data
were collected on. Model selection was undertaken using the method described in the
main text. From this, it was possible to calculate adjusted poverty scores for each
household sampled in the land use survey, based on the asset data collected.
156
Table B.1: Parameter estimates for the selected adjusted household poverty score GLMM, with a
the recalculated poverty score as the response. One random effect was included in the selected
model: village (SD= 0.532). Standard errors are shown in parentheses.
Variable
intercept
roof_type2
roof_type 3
roof_type 4
walls_type 1
walls_type 2
walls_type 3
walls_type 4
walls_type 5
owns_2cattle
owns_moto
owns_mobile
owns_dvdplayer
owns_strimmer
owns_wardrobe
Estimate
4.460 (2.192)
-1.791 (0.536)
-1.296 (0.437)
-0.334 (0.613)
3.121 (2.183)
4.040 (2.513)
3.092 (2.282)
4.543 (2.644)
3.476 (2.162)
0.162 (0.024)
0.724 (0.244)
0.978 (0.226)
1.338 (0.209)
0.643 (0.219)
0.805 (0.370)
157
t values
2.035
-3.341
-2.968
-0.545
1.430
1.608
1.355
1.719
1.608
6.665
2.969
4.323
6.400
2.928
2.176
Figure B.1: Map of household land use in Andong Kraloeng for the 43 households sampled. The area surrounding the residential and agricultural zones is
community NTFP forest in which all clearance is illegal.
158
Figure B.2: Map of household land use in O Rona for the 70 households sampled. The area outside of the village boundary is all under protection (either as part of
SPF or SWS) and, hence, clearance is restricted.
159
Table B.2: Variables used in statistical modelling.
Variable
log_area
log_ill_area
law_compliance
Data type
continuous
continuous
binary
Variable type
response
response
response
reg_compliance
binary
response
age
yrs_village
gender
indigenous
continuous
continuous
categorical
binary
explanatory
explanatory
explanatory
explanatory
AME
continuous
explanatory
hh_size
immigrant
community
position
labour
shop
employed
service
poverty_score
settlement
continuous
binary
binary
binary
binary
binary
binary
binary
continuous
categorical
explanatory
explanatory
explanatory
explanatory
explanatory
explanatory
explanatory
explanatory
explanatory
random
Description
ln(total area of land claimed by respondent household)
ln(total area of illegal land claimed by respondent household)
whether or not the respondent household was compliant with the 2001 Land Law
whether or not the respondent household was compliant with the five community
regulations considered
age of respondent
number of years respondent had lived in village
gender of respondent
whether the head of the respondent household was an IP
household adult male equivalence with recommended daily allowances taken from
Barba and Cabrera (2008)
total number of members in respondent household
whether or not the respondent household lived in the village at the time of ICT
whether or not the respondent household is a member of the village IC
whether or not the respondent held a formal position of authority
whether or not the respondent was involved in wage labour
whether or not the respondent household operated a shop
whether or not a member of respondent’s household was employed
whether or not the respondent’s household operated a village service
adjusted household poverty score of respondent household
sub-village settlement in which respondent household lived
160
Table B.3: Parameter estimates for the selected legal compliance GLMM, with a binary
compliance variable as the response. P values significant at the 95% confidence level are shown
in bold. One random effect was included in the selected model: settlement. The full list of
variables modelled, plus their explanations, are given in Appendix B; Table B.2.
Variable
intercept
community
ln(area + 0.125)
Estimate SE
-2.953
1.224
3.502
1.411
-1.345
0.432
P Values
0.016
0.013
0.002
Table B.4: Parameter estimates for the selected community regulation compliance GLMM, with
a binary compliance variable as the response. P values significant at the 95% confidence level are
shown in bold. One random effect was included in the selected model: settlement. The full list of
variables modelled, plus their explanations, are given in Appendix B; Table B.2.
Variable
intercept
community
ln(area + 0.125)
law_comp
Estimate SE
1.375
0.677
2.479
0.762
-1.685
0.432
-1.342
0.651
P Values
0.042
0.001
<< 0.000
0.039
References
Barba, C. & Cabrera, M. (2008) Recommended Dietary Allowances harmonization in
Southeast Asia. Asia Pacific Journal of Clinical Nutrition. 17 (S2), 405-408.
Davies, R. (1997) Beyond wealth ranking: The democratic definition and
measurement of poverty. Briefing note. ODI workshop "Indicators of Poverty:
Operational Significance", London.
161
Appendix C
Supplementary material for Chapter 5
Table C.1: Key village type selection criteria by village. Household data from Pollard & Evans
(2009). O Am household data (*) from the national commune database (NCDD, 2008). Bunong %
refers to the percentage of Bunong households within each village. Type gives the a priori
characterisation of each village.
Village
Chakchar
Gati
Kmoum
O Chrar
Rokathmei
Sre Andoal
Andong Kraleung
Pu Char
Pu Haim
Sre Levi
Sre Preah
O Am
O Rona
No. HHs
124
54
62
27
51
50
107
66
252
28
112
733*
160
Bunong %
87%
94%
100%
93%
92%
100%
95%
92%
90%
100%
66%
0%
46%
162
Km to market
60
22
55
18
40
48
25
14
19
10
12
0
8
Type
1
1
1
1
1
1
2
2
2
2
2
3
3
Sample size
n = 20
n = 20
n = 20
n = 20
n = 20
n = 20
n = 20
n = 20
n = 20
n = 20
n = 40
n = 80
n = 40
CPR game script
[Introductions]
Before I begin to explain what we are going to do today I would like to make a few
general comments. I will then explain the basic principles of the games that we are
going to play. After I have finished this explanation I will take you through four
simple examples and then we will play some practice games to help you to understand
the process.
We will be playing games for cash. Any money that you receive in the games is yours
to keep and take home. This will be given to you at the end of the day. How much
money you receive is dependent on how you play the games, and how other people
who are playing with you play. The money for these games has been supplied to
undertake scientific research. If at any time you find that this is something that you do
not wish to participate in for any reason, you are of course free to leave whether we
have started the game or not. If you choose to leave after you have played some of the
games, you will be paid the money that you have earned.
It is important that when you are sitting in the group like this you do not discuss the
game or how you think you should play the game. Please feel free to talk amongst
yourselves but do not discuss the games themselves. This can spoil the game for
everyone. If it happens we may not be able to play the game with you today.
[If this is the second or third day of the games within the village add the following:
“You are going to play a very different game than the games played before. If you
have heard comments about other games from people who have played games on
previous days, whatever they have mentioned might not apply today. If you follow
their advice you may end up earning less money than you could do so please just
listen to my explanation and play the game how you think best.”]
We will be playing three games today. We will then individually ask you some
questions about your thoughts on the day. We should be finished by about 5pm. If you
think you will not be able to stay that long please let us know now.
Each game follows the same basic principles which I will explain now. It is important
that you listen as closely as possible as you will need to understand how the games
work if you are to receive any money. We will run through a number of different
examples for you to help you understand how the games are played. If you have any
questions about how to play the games once they have been explained please put your
hand up and we will try to answer them.
I am now going to explain the game.
163
Explanation
In all the games we are going to play you will be in groups of 10 other people from
this village. In some games you will know who is in your group and in other games
you won’t. In each game there are 100 fish in a communal pond from which the
people in your group can harvest. You will be asked to individually decide on the
number of fish that you want to harvest. You can decide to harvest a maximum of 10
fish and a minimum of 0. For each fish that you harvest you will receive 100 KHR.
However, harvesting from the pond reduces the availability of fish for the village. The
number of fish left at the end of the season is an indication of the future value of the
fish to your group. At the end of each season the number of unharvested fish will be
counted. Everyone in your group will receive 12 KHR for every fish remaining. So, to
summarise, if a fish is unharvested everyone in your group benefits but if a fish is
harvested only the person who harvested it receives any benefit.
[Show poster to show value of fish to individual and fish in the village]
We will play each game for 5 seasons. The number of fish in the communal pond will
be the same for every season that we play no matter how the group behaved in
previous seasons.
You will be asked to signal your choice by circling the number of fish that you want
to harvest on the sheet of paper that you have in front of you. Please have a look at
this sheet of paper now. As you can see there is a row for each season. Each season
has an option for you to select a harvest of 0 to 10 fish. So if you want to harvest four
fish, circle four pictures. If you decide to harvest eight fish, circle eight pictures. If
you decide not to harvest, circle the zero. [Holds up example] The top row on your
sheet has been filled out for you as an example to show how to select your choice. In
this case you can see that a harvest of four fish has been selected. Once you have
made your choice for the season we will collect your paper so that we can calculate
the total harvest for the group. We will then tell you the total harvest for that season
and tell you your individual winnings for that season.
After we have played 5 seasons the game will end.
[Play two practice rounds as per the control]
You should all now have a completed practice round sheet. As you can see we have
filled out the two boxes at the end of the row. The number in the first box is the total
number of fish taken out of the pond by your group. So this is the number of fish
circled by each member of your group all added up together. The second number is
the money that you receive from the round. Now remember that this is a practice
round so the money in that box won’t go towards your earnings at the end of the day.
We calculate this number by giving you 100 KHR for each of the fish you have
circled. We then give you 12 KHR for each of fish left unharvested after every
member of the group has selected their individual harvest.
I will now take you through four simple examples to show you the range of behaviour
possible:
Example 1: All members take 0 fish – payoff = 1200 KHR
164
Example 2: All members take 10 fish – payoff = 1000 KHR
Example 3: 9 take 10 fish/1 takes 0 fish – payoff = 1120/120 KHR
Example 4: 1 takes 10 fish/9 take 0 fish – payoff = 2080/1080 KHR
Before we start to play the game for real I would just like to ask you some simple
questions to make sure that you understand how the game works and how to use the
payoff and answer sheets. Please circle the correct answer on this question sheet for
each question. For all questions, assume that the other people in the group take the
same number of fish no matter what you choose to do.
[Hand out question sheets. Go through each question, making sure that people look at
their own sheet only. Once everyone has answered all the questions, collect the sheets
and mark down a score on each. Then go through the answers]
1.
2.
3.
4.
5.
Do you earn more if you harvest 3 or 8 fish?
Do you earn more if you harvest 7 or 2 fish?
Do you earn more if the group total is 55 or 85?
Do you earn more if the group total is 22 or 62?
If the group total is 67, do other people earn more if you harvest 8 or 4 fish?
We should now be ready to play the first game. If anyone still has any questions,
please ask now. Remember if you feel that this is not something that you wish to take
part in you are free to leave at any time and any money that you have earned up to the
point you leave will be given to you.
Control
[Prepare the room for the control treatment. Arrange the desks in straight lines facing
the front of the room. Place a results sheet and desk number at each seat, making sure
to write the desk number on the results sheet. Invite the participants to sit down]
Ok, we are now ready to play the [game number] game. Let’s quickly run over the
details. You will be asked to decide how many fish you want to harvest each season
from the communal pond. You can decide to harvest a maximum of 10 fish and a
minimum of 0. For each fish that you decide to harvest you will receive 100 KHR.
This is the same for every member of your group. Everyone in the group benefits
from having fish remaining unharvested at the end of the season so that you will each
receive 12 KHR for each fish left in the communal pond.
This game is played in private, which means that you will not know who is in your
group. It is very important that you do not talk to each other during this game as this
will spoil the game for everybody. We will ask you to decide how many fish you want
to harvest for and to mark your decision on the sheet in front of you. We will then
calculate the total harvest for the group and your individual payoffs. Nobody else will
be told how much you have chosen to harvest or how much you have received. We
will play the game for 5 seasons which will probably last for about an hour.
We are now ready to start the game. From now on all the decisions that you make will
count towards the game. Please now decide how many fish you would like to harvest
in the first season and mark your decision on your sheet.
165
[Check that people only look at their own sheet. Help anyone who is having trouble.
Once everyone has marked their decision collect all the sheets and total the group
clearance. Then calculate each individual’s earnings and write both on each sheet.
Return the sheets to the respective participants]
As you can see, we have filled out the two boxes from that round. Remember the top
box shows you the amount cleared by the rest of your group and the lower box gives
you your earnings from that round.
[Repeat until all 5 seasons have been played making sure to say which number season
is currently being played when asking players to make their decision. Collect the
results sheets at the end of the game]
Peer-pressure
[Prepare the rooms for the single group treatments. Arrange all desks in a circle
facing each other. Place a results sheet on each desk and ask participants to sit in
ascending numerical order. Make sure that the correct number is on each sheet]
Ok, we are now ready to play the [game number] game. Let’s quickly run over the
details. You will be asked to decide how many fish you want to harvest each season
from the communal forest. You can decide to harvest a maximum of 10 fish and a
minimum of 0. For each fish you decide to harvest you will receive 100 KHR. This is
the same for every member of your group. Everyone in the group benefits from
having fish remaining unharvested at the end of the season so that you will each
receive 12 KHR for each fish left in the pond.
In this game you will know who is in your group and the number of days you choose
to harvest will not be private. Before each round you will be given a few minutes in
which you are free to discuss your decision and the decisions of the other players with
the rest of your group. You can also make agreements about how many fish you think
each individual should be allowed to harvest although these agreements are not
binding. Once this period is over we will ask you to individually decide how many
fish you want to harvest and to mark your decision on the sheet in front of you as
before. We will then calculate the total harvest for your group and your individual
payoffs. Once we have handed back your sheets we will ask each of you individually
to tell the other members of the group how much you chose to harvest.
We will repeat this game for 5 seasons. After each season you have a few minutes to
talk about the game with the other members of the group before beginning the next
season. Once all 5 seasons have been played, the game will end.
Ok, now we are ready to start the game. Please discuss amongst yourselves how many
fish you think each person should be allowed to harvest.
[Allow group discussion for as long as the participants choose to talk. Once the
discussion has stopped, ask the group whether or not they have reached a group
decision. Accept any uncontested answer and ask what decision has been reached. If
there is a disagreement about whether a decision has been reached, allow the
participants the opportunity to resolve the disagreement. If they resolve the
166
disagreement, ask what decision has been reached. If it is unresolved, note down the
values discussed]
Now you have finished your discussion, please decide how many fish you would like
to harvest in the first season and mark your decision on your sheet. Remember that the
group decision is not binding and you can choose any harvest from 0 to 10 fish.
[Check that people only look at their own sheet. Help anyone who is having trouble.
Once everyone has marked their decision collect all the sheets and total the group
clearance. Then calculate each individual’s earnings and write both on each sheet.
Return the sheets to the respective participants]
As you can see, we have filled out the two boxes from that round. Remember the top
box shows you the amount cleared by the rest of your group and the lower box gives
you your earnings from that round.
[Help each individual to read out their earnings for that season. Repeat until all 5
seasons have been played making sure to say which number season is currently being
played when asking players to make their decision. Select a different player each time
to begin declaring their decision for the season. Collect the results sheets and
feedback sheets at the end of the game.]
Weak enforcement
[Prepare the rooms for the single group treatments. Arrange all desks in a circle
facing each other. Place a results sheet on each desk and ask participants to sit in
ascending numerical order. Make sure that the correct number is on each sheet]
Ok, we are now ready to play the [game number] game. Let’s quickly run over the
details. You will be asked to decide how many fish you want to harvest each season
from the communal pond. You can decide to harvest a maximum of 10 fish and a
minimum of 0. For each fish that you decide to harvest you will receive 100 KHR.
This is the same for every member of your group. Everyone in the group benefits
from having fish remaining unharvested at the end of the season so that you will each
receive 12 KHR for each fish left in the communal pond.
In this game you will know who is in your group and the number of days you choose
to harvest will not be private. Before each round you will be given a few minutes in
which you are free to discuss your decision and the decisions of the other players with
the rest of your group. You can also make agreements about how much fish you think
each individual should be allowed to harvest although these agreements are not
binding. Once this period is over we will ask you to individually decide how many
fish you want to harvest and to mark your decision on the sheet in front of you as
before. We will then calculate the total harvest for your group and your individual
payoffs. Once we have handed back your sheets we will ask each of you individually
to tell the other members of the group how much you chose to harvest.
For this game, an external authority has made a rule that you are not allowed to
extract more than two fish from the pond in any single season. If you choose to fish,
there is a chance that your decision will be penalised. At the end of each season we
will roll a ten-sided dice, which will determine who in the group is monitored. There
167
will be a 1 in 10 chance that the external authority will monitor you. If you are
monitored and you have decided to do any fishing in the season you will be fined 125
KHR for each fish that you harvest over the permitted threshold of two fish. This
means that if, for example, you are monitored and have chosen to harvest four fish,
you will be fined 250 KHR (125 KHR for the two fish you harvested above the
threshold of two).
We will repeat this game for 5 seasons. After each season you have a few minutes to
talk about the game with the other members of the group before beginning the next
season. Once all 5 seasons have been played, the game will end.
Ok, now we are ready to start the game. Please discuss amongst yourselves how many
fish you think each person should be allowed to harvest.
[Allow group discussion for as long as the participants choose to talk. Once the
discussion has stopped, ask the group whether or not they have reached a group
decision. Accept any uncontested answer and ask what decision has been reached. If
there is a disagreement about whether a decision has been reached, allow the
participants the opportunity to resolve the disagreement. If they resolve the
disagreement, ask what decision has been reached. If it is unresolved, note down the
values discussed]
Now you have finished your discussion, please decide how many fish you would like
to harvest in the first season and mark your decision on your sheet. Remember that the
group decision is not binding and you can choose any harvest from 0 to 10 fish and
that if you harvest more than the two fish you are allowed to, you may be fined.
[Check that people only look at their own sheet. Help anyone who is having trouble.
Once everyone has marked their decision collect all the sheets and total the group
clearance. Then calculate each individual’s earnings and write both on each sheet.
Return the sheets to the respective participants]
As you can see, we have filled out the two boxes from that round. Remember the top
box shows you the amount cleared by the rest of your group and the lower box gives
you your earnings from that round.
[Help each individual to read out their earnings for that season. Ask each participant
to roll the die. If they roll a 1, they will be monitored and fined 125 KHR for each fish
they harvest above two. Correct their payoff and announce the new amount to the rest
of the group. Repeat until all 5 seasons have been played making sure to say which
number season is currently being played when asking players to make their decision.
Select a different player each time to begin declaring their decision for the season.
Collect the results sheets at the end of the game]
Strong enforcement
[Prepare the rooms for the single group treatments. Arrange all desks in a circle
facing each other. Place a results sheet on each desk and ask participants to sit in
ascending numerical order. Make sure that the correct number is on each sheet]
168
Ok, we are now ready to play the [game number] game. Let’s quickly run over the
details. You will be asked to decide how many fish you want to harvest each season
from the communal pond. You can decide to harvest a maximum of 10 fish and a
minimum of 0. For each fish that you decide to harvest you will receive 100 KHR.
This is the same for every member of your group. Everyone in the group benefits
from having fish remaining unharvested at the end of the season so that you will each
receive 12 KHR for each fish left in the communal pond.
In this game you will know who is in your group and the number of days you choose
to harvest will not be private. Before each round you will be given a few minutes in
which you are free to discuss your decision and the decisions of the other players with
the rest of your group. You can also make agreements about how much fish you think
each individual should be allowed to harvest although these agreements are not
binding. Once this period is over we will ask you to individually decide how many
fish you want to harvest and to mark your decision on the sheet in front of you as
before. We will then calculate the total harvest for your group and your individual
payoffs. Once we have handed back your sheets we will ask each of you individually
to tell the other members of the group how much you chose to harvest.
For this game, an external authority has made a rule that you are not allowed to
extract more than two fish from the pond in any single season. If you choose to fish,
there is a chance that your decision will be penalised. At the end of each season we
will roll a ten-sided dice, which will determine who in the group is monitored. There
will be a 4 in 10 chance that the external authority will monitor you. If you are
monitored and you have decided to do any fishing in the season you will be fined 125
KHR for each fish that you harvest over the permitted threshold of two fish. This
means that if, for example, you are monitored and have chosen to harvest four fish,
you will be fined 250 KHR (125 KHR for the two fish you harvested above the
threshold of two).
We will repeat this game for 5 seasons. After each season you have a few minutes to
talk about the game with the other members of the group before beginning the next
season. Once all 5 seasons have been played, the game will end.
Ok, now we are ready to start the game. Please discuss amongst yourselves how many
fish you think each person should be allowed to harvest.
[Allow group discussion for as long as the participants choose to talk. Once the
discussion has stopped, ask the group whether or not they have reached a group
decision. Accept any uncontested answer and ask what decision has been reached. If
there is a disagreement about whether a decision has been reached, allow the
participants the opportunity to resolve the disagreement. If they resolve the
disagreement, ask what decision has been reached. If it is unresolved, note down the
values discussed]
Now you have finished your discussion, please decide how many fish you would like
to harvest in the first season and mark your decision on your sheet. Remember that the
group decision is not binding and you can choose any harvest from 0 to 10 fish and
that if you harvest more than the two fish you are allowed to, you may be fined.
169
[Check that people only look at their own sheet. Help anyone who is having trouble.
Once everyone has marked their decision collect all the sheets and total the group
clearance. Then calculate each individual’s earnings and write both on each sheet.
Return the sheets to the respective participants]
As you can see, we have filled out the two boxes from that round. Remember the top
box shows you the amount cleared by the rest of your group and the lower box gives
you your earnings from that round.
[Help each individual to read out their earnings for that season. Ask each participant
to roll the die. If they roll a 1, 2, 3 or 4 they will be monitored and fined 125 KHR for
each fish they harvest above two. Correct their payoff and announce the new amount
to the rest of the group. Repeat until all 5 seasons have been played making sure to
say which number season is currently being played when asking players to make their
decision. Select a different player each time to begin declaring their decision for the
season. Collect the results sheets at the end of the game]
Individual payments
[Prepare the rooms for the single group treatments. Arrange all desks in a circle
facing each other. Place a results sheet on each desk and ask participants to sit in
ascending numerical order. Make sure that the correct number is on each sheet]
Ok, we are now ready to play the [game number] game. Let’s quickly run over the
details. You will be asked to decide how many fish you want to harvest each season
from the communal pond. You can decide to harvest a maximum of 10 fish and a
minimum of 0. For each fish that you decide to harvest you will receive 100 KHR.
This is the same for every member of your group. Everyone in the group benefits
from having fish remaining unharvested at the end of the season so that you will each
receive 12 KHR for each fish left in the communal pond.
In this game you will know who is in your group and the number of days you choose
to harvest will not be private. Before each round you will be given a few minutes in
which you are free to discuss your decision and the decisions of the other players with
the rest of your group. You can also make agreements about how much fish you think
each individual should be allowed to harvest although these agreements are not
binding. Once this period is over we will ask you to individually decide how many
fish you want to harvest and to mark your decision on the sheet in front of you as
before. We will then calculate the total harvest for your group and your individual
payoffs. Once we have handed back your sheets we will ask each of you individually
to tell the other members of the group how much you chose to harvest.
In addition, for this game there is an opportunity to earn a bonus payment from an
external organisation. These payments are designed to reduce the number of fish that
the group harvests. If you choose to harvest two fish or fewer each season, you will be
rewarded by a bonus payment of 250 KHR. This is irrespective of how other people in
your group behave. If you harvest more than two fish, you will not be eligible for the
bonus.
170
We will repeat this game for 5 seasons. After each season you have a few minutes to
talk about the game with the other members of the group before beginning the next
season. Once all 5 seasons have been played, the game will end.
Ok, now we are ready to start the game. Please discuss amongst yourselves how many
fish you think each person should be allowed to harvest.
[Allow group discussion for as long as the participants choose to talk. Once the
discussion has stopped, ask the group whether or not they have reached a group
decision. Accept any uncontested answer and ask what decision has been reached. If
there is a disagreement about whether a decision has been reached, allow the
participants the opportunity to resolve the disagreement. If they resolve the
disagreement, ask what decision has been reached. If it is unresolved, note down the
values discussed]
Now you have finished your discussion, please decide how many fish you would like
to harvest in the first season and mark your decision on your sheet. Remember that the
group decision is not binding and you can choose any harvest from 0 to 10 fish.
Remember also that in this game if you choose to harvest two fish or fewer you will
be rewarded with a bonus payment of 250 KHR.
[Check that people only look at their own sheet. Help anyone who is having trouble.
Once everyone has marked their decision collect all the sheets and total the group
clearance. Then calculate each individual’s earnings and write both on each sheet.
Return the sheets to the respective participants]
As you can see, we have filled out the two boxes from that round. Remember the top
box shows you the amount cleared by the rest of your group and the lower box gives
you your earnings from that round.
[Help each individual to read out their earnings for that season. Repeat until all 5
seasons have been played making sure to say which number season is currently being
played when asking players to make their decision. Select a different player each time
to begin declaring their decision for the season. Collect the results sheets at the end of
the game.]
Individual payments with weak enforcement
[Prepare the rooms for the single group treatments. Arrange all desks in a circle
facing each other. Place a results sheet on each desk and ask participants to sit in
ascending numerical order. Make sure that the correct number is on each sheet]
Ok, we are now ready to play the [game number] game. Let’s quickly run over the
details. You will be asked to decide how many fish you want to harvest each season
from the communal pond. You can decide to harvest a maximum of 10 fish and a
minimum of 0. For each fish that you decide to harvest you will receive 100 KHR.
This is the same for every member of your group. Everyone in the group benefits
from having fish remaining unharvested at the end of the season so that you will each
receive 12 KHR for each fish left in the communal pond.
171
In this game you will know who is in your group and the number of days you choose
to harvest will not be private. Before each round you will be given a few minutes in
which you are free to discuss your decision and the decisions of the other players with
the rest of your group. You can also make agreements about how much fish you think
each individual should be allowed to harvest although these agreements are not
binding. Once this period is over we will ask you to individually decide how many
fish you want to harvest and to mark your decision on the sheet in front of you as
before. We will then calculate the total harvest for your group and your individual
payoffs. Once we have handed back your sheets we will ask each of you individually
to tell the other members of the group how much you chose to harvest.
For this game, an external authority has made a rule that you are not allowed to
extract more than two fish from the pond in any single season. If you choose to fish,
there is a chance that your decision will be penalised. At the end of each season we
will roll a ten-sided dice, which will determine who in the group is monitored. There
will be a 1 in 10 chance that the external authority will monitor you. If you are
monitored and you have decided to do any fishing in the season you will be fined 125
KHR for each fish that you harvest over the permitted threshold of two fish. This
means that if, for example, you are monitored and have chosen to harvest four fish,
you will be fined 250 KHR (125 KHR for the two fish you harvested above the
threshold of two).
In addition, for this game there is an opportunity to earn a bonus payment from an
external organisation. These payments are designed to reduce the number of fish that
the group harvests. If you choose to harvest two fish or fewer each season, you will be
rewarded by a bonus payment of 250 KHR. This is irrespective of how other people in
your group behave. If you harvest more than two fish, you will not be eligible for the
bonus.
We will repeat this game for 5 seasons. After each season you have a few minutes to
talk about the game with the other members of the group before beginning the next
season. Once all 5 seasons have been played, the game will end.
Ok, now we are ready to start the game. Please discuss amongst yourselves how many
fish you think each person should be allowed to harvest.
[Allow group discussion for as long as the participants choose to talk. Once the
discussion has stopped, ask the group whether or not they have reached a group
decision. Accept any uncontested answer and ask what decision has been reached. If
there is a disagreement about whether a decision has been reached, allow the
participants the opportunity to resolve the disagreement. If they resolve the
disagreement, ask what decision has been reached. If it is unresolved, note down the
values discussed]
Now you have finished your discussion, please decide how many fish you would like
to harvest in the first season and mark your decision on your sheet. Remember that the
group decision is not binding and you can choose any harvest from 0 to 10 fish.
Remember also that in this game if you choose to harvest two fish or fewer you will
be rewarded with a bonus payment of 250 KHR and that if you harvest more than the
two fish you are allowed to, you may be fined.
172
[Check that people only look at their own sheet. Help anyone who is having trouble.
Once everyone has marked their decision collect all the sheets and total the group
clearance. Then calculate each individual’s earnings and write both on each sheet.
Return the sheets to the respective participants]
As you can see, we have filled out the two boxes from that round. Remember the top
box shows you the amount cleared by the rest of your group and the lower box gives
you your earnings from that round.
[Help each individual to read out their earnings for that season. Ask each participant
to roll the die. If they roll a 1 they will be monitored and fined 125 KHR for each fish
they harvest above two. Correct their payoff and announce the new amount to the rest
of the group. Repeat until all 5 seasons have been played making sure to say which
number season is currently being played when asking players to make their decision.
Select a different player each time to begin declaring their decision for the season.
Collect the results sheets at the end of the game.]
Individual payments with strong enforcement
[Prepare the rooms for the single group treatments. Arrange all desks in a circle
facing each other. Place a results sheet on each desk and ask participants to sit in
ascending numerical order. Make sure that the correct number is on each sheet]
Ok, we are now ready to play the [game number] game. Let’s quickly run over the
details. You will be asked to decide how many fish you want to harvest each season
from the communal pond. You can decide to harvest a maximum of 10 fish and a
minimum of 0. For each fish that you decide to harvest you will receive 100 KHR.
This is the same for every member of your group. Everyone in the group benefits
from having fish remaining unharvested at the end of the season so that you will each
receive 12 KHR for each fish left in the communal pond.
In this game you will know who is in your group and the number of days you choose
to harvest will not be private. Before each round you will be given a few minutes in
which you are free to discuss your decision and the decisions of the other players with
the rest of your group. You can also make agreements about how much fish you think
each individual should be allowed to harvest although these agreements are not
binding. Once this period is over we will ask you to individually decide how many
fish you want to harvest and to mark your decision on the sheet in front of you as
before. We will then calculate the total harvest for your group and your individual
payoffs. Once we have handed back your sheets we will ask each of you individually
to tell the other members of the group how much you chose to harvest.
For this game, an external authority has made a rule that you are not allowed to
extract more than two fish from the pond in any single season. If you choose to fish,
there is a chance that your decision will be penalised. At the end of each season we
will roll a ten-sided dice, which will determine who in the group is monitored. There
will be a 4 in 10 chance that the external authority will monitor you. If you are
monitored and you have decided to do any fishing in the season you will be fined 125
KHR for each fish that you harvest over the permitted threshold of two fish. This
means that if, for example, you are monitored and have chosen to harvest four fish,
you will be fined 250 KHR (125 KHR for the two fish you harvested above the
173
threshold of two).
In addition, for this game there is an opportunity to earn a bonus payment from an
external organisation. These payments are designed to reduce the number of fish that
the group harvests. If you choose to harvest two fish or fewer each season, you will be
rewarded by a bonus payment of 250 KHR. This is irrespective of how other people in
your group behave. If you harvest more than two fish, you will not be eligible for the
bonus.
We will repeat this game for 5 seasons. After each season you have a few minutes to
talk about the game with the other members of the group before beginning the next
season. Once all 5 seasons have been played, the game will end.
Ok, now we are ready to start the game. Please discuss amongst yourselves how many
fish you think each person should be allowed to harvest.
[Allow group discussion for as long as the participants choose to talk. Once the
discussion has stopped, ask the group whether or not they have reached a group
decision. Accept any uncontested answer and ask what decision has been reached. If
there is a disagreement about whether a decision has been reached, allow the
participants the opportunity to resolve the disagreement. If they resolve the
disagreement, ask what decision has been reached. If it is unresolved, note down the
values discussed]
Now you have finished your discussion, please decide how many fish you would like
to harvest in the first season and mark your decision on your sheet. Remember that the
group decision is not binding and you can choose any harvest from 0 to 10 fish.
Remember also that in this game if you choose to harvest two fish or fewer you will
be rewarded with a bonus payment of 250 KHR and that if you harvest more than the
two fish you are allowed to, you may be fined.
[Check that people only look at their own sheet. Help anyone who is having trouble.
Once everyone has marked their decision collect all the sheets and total the group
clearance. Then calculate each individual’s earnings and write both on each sheet.
Return the sheets to the respective participants]
As you can see, we have filled out the two boxes from that round. Remember the top
box shows you the amount cleared by the rest of your group and the lower box gives
you your earnings from that round.
[Help each individual to read out their earnings for that season. Ask each participant
to roll the die. If they roll a 1, 2, 3 or 4 they will be monitored and fined 125 KHR for
each fish they harvest above two. Correct their payoff and announce the new amount
to the rest of the group. Repeat until all 5 seasons have been played making sure to
say which number season is currently being played when asking players to make their
decision. Select a different player each time to begin declaring their decision for the
season. Collect the results sheets at the end of the game.]
174
Collective payments
[Prepare the rooms for the single group treatments. Arrange all desks in a circle
facing each other. Place a results sheet on each desk and ask participants to sit in
ascending numerical order. Make sure that the correct number is on each sheet]
Ok, we are now ready to play the [game number] game. Let’s quickly run over the
details. You will be asked to decide how many fish you want to harvest each season
from the communal pond. You can decide to harvest a maximum of 10 fish and a
minimum of 0. For each fish that you decide to harvest you will receive 100 KHR.
This is the same for every member of your group. Everyone in the group benefits
from having fish remaining unharvested at the end of the season so that you will each
receive 12 KHR for each fish left in the communal pond.
In this game you will know who is in your group and the number of days you choose
to harvest will not be private. Before each round you will be given a few minutes in
which you are free to discuss your decision and the decisions of the other players with
the rest of your group. You can also make agreements about how much fish you think
each individual should be allowed to harvest although these agreements are not
binding. Once this period is over we will ask you to individually decide how many
fish you want to harvest and to mark your decision on the sheet in front of you as
before. We will then calculate the total harvest for your group and your individual
payoffs. Once we have handed back your sheets we will ask each of you individually
to tell the other members of the group how much you chose to harvest.
For this game, an external authority has made a rule that you are not allowed to
extract more than two fish from the pond in any single season. If you choose to fish,
there is a chance that your decision will be penalised. At the end of each season we
will roll a ten-sided dice, which will determine who in the group is monitored. There
will be a 1 in 10 chance that the external authority will monitor you. If you are
monitored and you have decided to do any fishing in the season you will be fined 125
KHR for each fish that you harvest over the permitted threshold of two fish. This
means that if, for example, you are monitored and have chosen to harvest four fish,
you will be fined 250 KHR (125 KHR for the two fish you harvested above the
threshold of two).
In addition, for this game there is an opportunity to earn a bonus payment from an
external organisation. These payments are designed to reduce the number of fish that
the group harvests. Whether you receive the bonus is dependent on the total group
harvest. If at the end of the season the total harvest is equal or less than 20 fish,
everyone in the group will receive a bonus payment of 250 KHR. If the total group
harvest is more than 20 fish then nobody will receive a bonus payment. So for
example if you only harvest a single fish but the total group harvest is 27 fish then
you will not receive a bonus payment, but if you were to harvest four fish and total
group harvest was 18, say, then you would receive a bonus.
We will repeat this game for 5 seasons. After each season you have a few minutes to
talk about the game with the other members of the group before beginning the next
season. Once all 5 seasons have been played, the game will end.
175
Ok, now we are ready to start the game. Please discuss amongst yourselves how many
fish you think each person should be allowed to harvest.
[Allow group discussion for as long as the participants choose to talk. Once the
discussion has stopped, ask the group whether or not they have reached a group
decision. Accept any uncontested answer and ask what decision has been reached. If
there is a disagreement about whether a decision has been reached, allow the
participants the opportunity to resolve the disagreement. If they resolve the
disagreement, ask what decision has been reached. If it is unresolved, note down the
values discussed]
Now you have finished your discussion, please decide how many fish you would like
to harvest in the first season and mark your decision on your sheet. Remember that the
group decision is not binding and you can choose any harvest from 0 to 10 fish.
Remember also that in this game if the total group harvest is 20 fish or fewer
everyone in the group will be rewarded with a bonus payment of 250 KHR.
[Check that people only look at their own sheet. Help anyone who is having trouble.
Once everyone has marked their decision collect all the sheets and total the group
clearance. Then calculate each individual’s earnings and write both on each sheet.
Return the sheets to the respective participants]
As you can see, we have filled out the two boxes from that round. Remember the top
box shows you the amount cleared by the rest of your group and the lower box gives
you your earnings from that round.
[Help each individual to read out their earnings for that season. Repeat until all 5
seasons have been played making sure to say which number season is currently being
played when asking players to make their decision. Select a different player each time
to begin declaring their decision for the season. Collect the results sheets at the end of
the game.]
Collective payments with weak enforcement
[Prepare the rooms for the single group treatments. Arrange all desks in a circle
facing each other. Place a results sheet on each desk and ask participants to sit in
ascending numerical order. Make sure that the correct number is on each sheet]
Ok, we are now ready to play the [game number] game. Let’s quickly run over the
details. You will be asked to decide how many fish you want to harvest each season
from the communal pond. You can decide to harvest a maximum of 10 fish and a
minimum of 0. For each fish that you decide to harvest you will receive 100 KHR.
This is the same for every member of your group. Everyone in the group benefits
from having fish remaining unharvested at the end of the season so that you will each
receive 12 KHR for each fish left in the communal pond.
In this game you will know who is in your group and the number of days you choose
to harvest will not be private. Before each round you will be given a few minutes in
which you are free to discuss your decision and the decisions of the other players with
the rest of your group. You can also make agreements about how much fish you think
each individual should be allowed to harvest although these agreements are not
176
binding. Once this period is over we will ask you to individually decide how many
fish you want to harvest and to mark your decision on the sheet in front of you as
before. We will then calculate the total harvest for your group and your individual
payoffs. Once we have handed back your sheets we will ask each of you individually
to tell the other members of the group how much you chose to harvest.
For this game, an external authority has made a rule that you are not allowed to
extract more than two fish from the pond in any single season. If you choose to fish,
there is a chance that your decision will be penalised. At the end of each season we
will roll a ten-sided dice, which will determine who in the group is monitored. There
will be a 4 in 10 chance that the external authority will monitor you. If you are
monitored and you have decided to do any fishing in the season you will be fined 125
KHR for each fish that you harvest over the permitted threshold of two fish. This
means that if, for example, you are monitored and have chosen to harvest four fish,
you will be fined 250 KHR (125 KHR for the two fish you harvested above the
threshold of two).
In addition, for this game there is an opportunity to earn a bonus payment from an
external organisation. These payments are designed to reduce the number of fish that
the group harvests. Whether you receive the bonus is dependent on the total group
harvest. If at the end of the season the total harvest is equal or less than 20 fish,
everyone in the group will receive a bonus payment of 250 KHR. If the total group
harvest is more than 20 fish then nobody will receive a bonus payment. So for
example if you only harvest a single fish but the total group harvest is 27 fish then
you will not receive a bonus payment, but if you were to harvest four fish and total
group harvest was 18, say, then you would receive a bonus.
We will repeat this game for 5 seasons. After each season you have a few minutes to
talk about the game with the other members of the group before beginning the next
season. Once all 5 seasons have been played, the game will end.
Ok, now we are ready to start the game. Please discuss amongst yourselves how many
fish you think each person should be allowed to harvest.
[Allow group discussion for as long as the participants choose to talk. Once the
discussion has stopped, ask the group whether or not they have reached a group
decision. Accept any uncontested answer and ask what decision has been reached. If
there is a disagreement about whether a decision has been reached, allow the
participants the opportunity to resolve the disagreement. If they resolve the
disagreement, ask what decision has been reached. If it is unresolved, note down the
values discussed]
Now you have finished your discussion, please decide how many fish you would like
to harvest in the first season and mark your decision on your sheet. Remember that the
group decision is not binding and you can choose any harvest from 0 to 10 fish.
Remember also that in this game if the total group harvest is 20 fish or fewer
everyone in the group will be rewarded with a bonus payment of 250 KHR and that if
you harvest more than the two fish you are allowed to, you may be fined.
[Check that people only look at their own sheet. Help anyone who is having trouble.
Once everyone has marked their decision collect all the sheets and total the group
177
clearance. Then calculate each individual’s earnings and write both on each sheet.
Return the sheets to the respective participants]
As you can see, we have filled out the two boxes from that round. Remember the top
box shows you the amount cleared by the rest of your group and the lower box gives
you your earnings from that round.
[Help each individual to read out their earnings for that season. Ask each participant
to roll the die. If they roll a 1 they will be monitored and fined 125 KHR for each fish
they harvest above two. Correct their payoff and announce the new amount to the rest
of the group. Repeat until all 5 seasons have been played making sure to say which
number season is currently being played when asking players to make their decision.
Select a different player each time to begin declaring their decision for the season.
Collect the results sheets at the end of the game.]
Collective payments with strong enforcement
[Prepare the rooms for the single group treatments. Arrange all desks in a circle
facing each other. Place a results sheet on each desk and ask participants to sit in
ascending numerical order. Make sure that the correct number is on each sheet]
Ok, we are now ready to play the [game number] game. Let’s quickly run over the
details. You will be asked to decide how many fish you want to harvest each season
from the communal pond. You can decide to harvest a maximum of 10 fish and a
minimum of 0. For each fish that you decide to harvest you will receive 100 KHR.
This is the same for every member of your group. Everyone in the group benefits
from having fish remaining unharvested at the end of the season so that you will each
receive 12 KHR for each fish left in the communal pond.
In this game you will know who is in your group and the number of days you choose
to harvest will not be private. Before each round you will be given a few minutes in
which you are free to discuss your decision and the decisions of the other players with
the rest of your group. You can also make agreements about how much fish you think
each individual should be allowed to harvest although these agreements are not
binding. Once this period is over we will ask you to individually decide how many
fish you want to harvest and to mark your decision on the sheet in front of you as
before. We will then calculate the total harvest for your group and your individual
payoffs. Once we have handed back your sheets we will ask each of you individually
to tell the other members of the group how much you chose to harvest.
In addition, for this game there is an opportunity to earn a bonus payment from an
external organisation. These payments are designed to reduce the number of fish that
the group harvests. Whether you receive the bonus is dependent on the total group
harvest. If at the end of the season the total harvest is equal or less than 20 fish,
everyone in the group will receive a bonus payment of 250 KHR. If the total group
harvest is more than 20 fish then nobody will receive a bonus payment. So for
example if you only harvest a single fish but the total group harvest is 27 fish then
you will not receive a bonus payment, but if you were to harvest four fish and total
group harvest was 18, say, then you would receive a bonus.
178
We will repeat this game for 5 seasons. After each season you have a few minutes to
talk about the game with the other members of the group before beginning the next
season. Once all 5 seasons have been played, the game will end.
Ok, now we are ready to start the game. Please discuss amongst yourselves how many
fish you think each person should be allowed to harvest.
[Allow group discussion for as long as the participants choose to talk. Once the
discussion has stopped, ask the group whether or not they have reached a group
decision. Accept any uncontested answer and ask what decision has been reached. If
there is a disagreement about whether a decision has been reached, allow the
participants the opportunity to resolve the disagreement. If they resolve the
disagreement, ask what decision has been reached. If it is unresolved, note down the
values discussed]
Now you have finished your discussion, please decide how many fish you would like
to harvest in the first season and mark your decision on your sheet. Remember that the
group decision is not binding and you can choose any harvest from 0 to 10 fish.
Remember also that in this game if the total group harvest is 20 fish or fewer
everyone in the group will be rewarded with a bonus payment of 250 KHR and that if
you harvest more than the two fish you are allowed to, you may be fined.
[Check that people only look at their own sheet. Help anyone who is having trouble.
Once everyone has marked their decision collect all the sheets and total the group
clearance. Then calculate each individual’s earnings and write both on each sheet.
Return the sheets to the respective participants]
As you can see, we have filled out the two boxes from that round. Remember the top
box shows you the amount cleared by the rest of your group and the lower box gives
you your earnings from that round.
[Help each individual to read out their earnings for that season. Ask each participant
to roll the die. If they roll a 1, 2, 3 or 4 they will be monitored and fined 125 KHR for
each fish they harvest above two. Correct their payoff and announce the new amount
to the rest of the group. Repeat until all 5 seasons have been played making sure to
say which number season is currently being played when asking players to make their
decision. Select a different player each time to begin declaring their decision for the
season. Collect the results sheets at the end of the game.]
179
Decision sheet
0
1
2
3
4
5
6
7
8
9
10
___________________________________________________________________________________
0
0
0
0
0
1
2
3
4
5
6
7
8
9
10
1
2
3
4
5
6
7
8
9
10
1
2
3
4
5
6
7
8
9
10
1
2
3
4
5
6
7
8
9
10
1
2
3
4
5
6
7
8
9
10
180
CPR treatment payoff equations
Enforcement treatments (λ = 0.1 or 0.4):
100𝑥𝑗 + 12 ( 100 − ∑
𝜋𝑗 =
{
100𝑥𝑗 + 12 ( 100 − ∑
𝑛
𝑥𝑖 )
if 𝑥𝑗 ≤ 2
𝑖=1
𝑛
𝑥𝑖 ) − 125λ
otherwise
𝑥𝑖 ) + 250
if 𝑥𝑗 ≤ 2
𝑖=1
Individual payment treatment:
𝜋𝑗 =
100𝑥𝑗 + 12 ( 100 − ∑
𝑛
𝑖=1
𝑛
100𝑥𝑗 + 12 ( 100 − ∑ 𝑥𝑖 )
{
𝑖=1
otherwise
Individual payment plus enforcement treatments (λ = 0.1 or 0.4):
100𝑥𝑗 + 12 ( 100 − ∑
𝜋𝑗 =
{
100𝑥𝑗 + 12 ( 100 − ∑
𝑛
𝑥𝑖 ) + 250
if 𝑥𝑗 ≤ 2
𝑥𝑖 ) − 125λ
otherwise
𝑖=1
𝑛
𝑖=1
Collective payments treatment:
100𝑥𝑗 + 12 ( 100 − ∑
𝜋𝑗 =
{
𝑛
𝑥𝑖 ) + 250
𝑖=1
𝑛
100𝑥𝑗 + 12 ( 100 − ∑
if ∑
𝑥𝑖 )
𝑛
𝑥𝑖 ≤ 20
𝑖=1
otherwise
𝑖=1
Collective payment plus enforcement treatments (λ = 0.1 or 0.4):
100𝑥𝑗 + 12 ( 100 − ∑
𝜋𝑗 =
{
𝑛
𝑥𝑖 ) + 250
𝑖=1
𝑛
if ∑
𝑛
𝑥𝑖 ≤ 20, 𝑥𝑗 ≤ 2
𝑖=1
𝑛
100𝑥𝑗 + 12 ( 100 − ∑
𝑥𝑖 ) + 250 − 125𝜆 if ∑
100𝑥𝑗 + 12 ( 100 − ∑
𝑥𝑖 )
if ∑
𝑥𝑖 > 20, 𝑥𝑗 ≤ 2
𝑥𝑖 ) − 125𝜆
if ∑
𝑥𝑖 > 20, 𝑥𝑗 > 2
𝑖=1
𝑛
𝑖=1
𝑛
100𝑥𝑗 + 12 ( 100 − ∑
𝑖=1
181
𝑥𝑖 ≤ 20, 𝑥𝑗 > 2
𝑖=1
𝑛
𝑖=1
𝑛
𝑖=1
Table C.2: Variables used in statistical modelling of the CPR game.
Variable
fish
treatment
previous
age
yrs_village
education
literacy
gender
head
ethnicity
dependency
gender_hhh
understanding
numeracy
position
decision
livelihood
round
game
day
village
session
group
individual
Data type
bounded proportional
categorical
categorical
continuous
continuous
continuous
binary
categorical
binary
categorical
continuous
categorical
bounded proportional
categorical
binary
binary
continuous
categorical
categorical
categorical
categorical
categorical
categorical
categorical
Variable type
response
explanatory
explanatory
explanatory
explanatory
explanatory
explanatory
explanatory
explanatory
explanatory
explanatory
explanatory
explanatory
explanatory
explanatory
explanatory
explanatory
explanatory
explanatory
explanatory
random effect
random effect
random effect
random effect
Description
number of fish taken by one subject
experimental treatment played
experimental treatment played previously
age of subject
number of years subject had lived in village
number of years in education
whether or not subject was literate
gender of subject
whether or not the subject was the head of family
ethnicity of subject
dependency ratio of household
gender of household head
number of correct answers to understanding questions
scored ability to correctly identify and order series of numbers
whether or not the subject held a formal position of authority
whether or not a group decision was made
number of livelihood activities followed by household
number of the round played
when in the day the observation was recorded
which day in the village the observation was recorded
village in which the observation was recorded
day on which the observation was recorded
group from which the observation was recorded
subject from which the observation was recorded
182
Table C.3: Parameter estimates for the selected CPR game GLMM, with the number of fish
harvested in a round as the response variable. The difference in the number of fish taken was
found by calculating the expected number of fish taken for each variable and comparing this
with the number expected for the intercept. P values significant at the 95% confidence level are
shown in bold. Three random effects were included in the model: individual (SD = 0.484), group
(SD = 0.561) and session number (SD = 0.591). The full list of variables modelled, plus their
explanations, are given in Supplementary Materials Table C.2.
Variable
intercept
private
weak enforcement
strong enforcement
individual payments
collective payments
ind. w. weak enf.
ind. w. strong enf.
coll. w. weak enf.
coll. w. strong enf.
education
decision
practice
round 2
round 3
round 4
round 5
Estimate
-0.431
-0.044
-0.041
0.158
-1.408
-1.158
-0.948
-0.506
-0.855
-1.164
0.024
-0.328
0.079
0.349
0.237
0.410
0.340
SE
0.296
0.265
0.343
0.343
0.326
0.329
0.326
0.329
0.345
0.341
0.009
0.084
0.016
0.032
0.032
0.032
0.032
No. of fish
3.938
3.833
3.841
4.321
1.371
1.695
2.011
2.815
2.164
1.686
3.994
3.188
4.129
4.793
4.515
4.947
4.773
Diff in no. fish
-0.105
-0.097
+0.383
-2.567
-2.243
-1.927
-1.123
-1.774
-2.252
+0.056
-0.750
+0.191
+0.855
+0.577
+1.009
+0.835
P values
0.145
0.868
0.905
0.644
<< 0.001
0.001
0.004
0.124
0.013
< 0.001
0.009
<< 0.001
<< 0.001
<< 0.001
<< 0.001
<< 0.001
<< 0.001
Table C.4: Comparison of the predicted difference in the number of fish taken by an individual
for different treatments between round 1 (Table 5.2) and round 4 (the treatment for which the
estimated effect is most different from round 1).
Treatment
intercept (peer pressure)
private
weak enforcement
strong enforcement
individual payments
collective payments
ind. w. weak enf.
ind. w. strong enf.
coll. w. weak enf.
coll. w. strong enf.
Difference in fish taken
in round 1
-0.11
-0.10
+0.38
-2.57
-2.24
-1.93
-1.12
-1.77
-2.25
183
Difference in fish taken in
round 4
-0.11
-0.10
0.40
-3.02
-2.60
-2.20
-1.23
-2.01
-2.61
Figure C.1: Average individual extraction per round during the enforcement treatments of the CPR game
184
Figure C.2: Average individual extraction per round during the individual payments treatments of the CPR game.
185
Figure C.3: Average individual extraction per round during the collective payments treatments of the CPR game.
186
Appendix D
Supplementary material for Chapter 6
Scenarios script
Interview Introduction
[Introduce research team]
I would like you to think about the future and what you would do if different things
changed. I will present different possible futures to you and I would like you to
imagine how you and your household would respond to the conditions in those
scenarios. Each scenario will cover a period of five years and only one condition will
be different to now.
For example, I might present you with a scenario of the future in which the price of
cassava has increased to double its current value but everything else was the same as
it is now. I would then want you to think about how that would affect you and the rest
of your household. For, instance if the price of cassava was double what you are paid
now, could you afford to hire more labour? Would you want to change the area of
land that you own by, for example, clearing or buying more land, or even selling your
land? Would you change the proportion of your land on which you grow cassava?
Would you change the proportion of your labour that you devote to farming as
opposed to other activities, such as resin collection? How do you think it would
change your relationship with other families in the village?
Or maybe you would not do anything different to what you do currently. If that was
the case, would there be anything stopping you from making changes? For example,
would it be possible for you to hire more labour than there is available to you now? Is
it possible to buy/sell or clear more land?
So what I am asking you to do is to think about how you might react to changes in the
future.
187
Scenarios
Cassava Price Increase
I want you to imagine that as, with our example, the price at which you can sell your
cassava doubles. This means that the cassava trader pays you twice as much as they
pay you this year. Imagine that this change continues for five years - how do you
think this will affect you? Remember in this future, everything else will remain as it is
now.
Cassava Price Decrease
I want you to imagine that the price at which you can sell your cassava halves. This
means that the cassava trader pays you half as much as they pay you this year.
Imagine that this change continues for five years - how do you think this will affect
you? Remember in this future, everything else will remain as it is now.
Law Enforcement Effort Increase
I want you to imagine that the protected area authorities, the Forestry Administration,
increase the number of law enforcement patrols targeting forest clearance and fields
located within the protected area. This extra effort means that the patrols are four
times more likely to catch illegal land clearance. Imagine that this change continues
for five years - how do you think this will affect you? Remember in this future,
everything else will remain as it is now.
Individual Payments to Reduce Forest Clearance
Now, I want you to imagine that the protected area authorities are offering another
type of payments to your village for people not to clear forest. This payment will be
worth $200 to everyone who receives it and if you receive it you will be given money
directly. Imagine you will receive a payment if you do not clear forest around your
fields or the village land. This payment is only dependent on your own behaviour.
That means that if you do not clear forest you will receive the payment. Imagine that
this change continues for five years - how do you think this will affect you?
Remember in this future, everything else will remain as it is now.
188
Communal Payments to Reduce Forest Clearance
Now, I want you to imagine that the protected area authorities are offering payments
to your village for people not to clear forest. This payment will be worth $200 to
everyone who receives it and if you receive it you will be given money directly.
Imagine you and the other members of your krom will receive a payment if there is no
forest clearance around your fields or the village land. This payment is dependent on
the behaviour of everybody, not just you. That means that if other people from your
krom clear but you do not, you will not receive the payment. Similarly, if other people
from your krom did not clear but you did, that would mean that nobody would receive
the payment. Only if nobody in your krom, including you, did not clear would
everybody receive the payment. Imagine that this change continues for five years how do you think this will affect you? Remember in this future, everything else will
remain as it is now.
Payment to a Village Fund
Finally, I want you to imagine that protected area authorities are again offering
payments to your village for people not to clear forest. This time the payments will be
made into a village fund to pay for improved village infrastructure, such as roads,
drainage, schools or health clinics. The money paid into this fund will be dependent
on the village on there being no forest clearance by anyone in the village. If people
choose to clear forest, there will be less money paid into the village fund and it will
take longer to pay for improvements to village infrastructure. Imagine that this change
continues for five years - how do you think this will affect you? Remember in this
future, everything else will remain as it is now.
189
Table D.1: Variables used in statistical modelling in Chapter 6.
Variable
pro
reduced
scenario
age
log_area
yrs_village
education
gender
indigenous
hh_size
labour
shop
employed
service
poverty_score
livelihood_zone
individual
Data type
binary
binary
categorical
continuous
continuous
continuous
continuous
categorical
binary
continuous
binary
binary
binary
binary
continuous
categorical
categorical
Variable type
response
response
explanatory
explanatory
explanatory
explanatory
explanatory
explanatory
explanatory
explanatory
explanatory
explanatory
explanatory
explanatory
explanatory
explanatory
random effect
Description
whether or not respondent acted in a pro-conservation manner
whether or not the respondent reduced clearance relative to baseline
scenario for which behaviour was predicted
age of respondent
natural logarithm of area of productive land held by household
number of years respondent had lived in village
number of years in education
gender of respondent
whether or not the respondent belonged to IP
number of members of household
whether or not the respondent household was involved in wage labour
whether or not the respondent household operated a shop
whether or not a member of respondent’s household was employed
whether or not the respondent’s household operated a village service
household poverty score
livelihood zone in which village was situated
subject from which the observation was recorded
190